A method, system and device for locating a broadband oscillation source of an offshore wind farm

Abstract

This invention discloses a method, system, and equipment for locating broadband oscillation sources in offshore wind farms. The method includes: synchronously acquiring three-phase voltage signals from the wind farm's common junction point and each feeder branch; extracting and clustering spectral peaks; adaptively determining the number of modes and the initial center frequency of each mode in variational mode decomposition; using a variational mode decomposition algorithm to decompose the original voltage and current signals into voltage mode components and current mode components; calculating the cumulative interactive energy flow sequence for each mode component; selecting a time window segment for linear regression to obtain the energy flow trend slope; determining the source-load nature of nodes by the positive or negative sign of the energy flow trend slope; calculating the oscillation contribution based on the proportion of the slope amplitude of multiple nodes; and finally locating the dominant oscillation source. This invention, by replacing the traditional instantaneous value judgment with the trend slope, can effectively eliminate random noise interference and the influence of filter phase lag, thereby achieving accurate location of oscillation sources under complex operating conditions.

Description

Technical Field

[0001] This invention relates to the field of power system oscillation source location technology, specifically to a broadband oscillation source location method, system, and equipment for offshore wind farms. Background Technology

[0002] With the large-scale development of offshore wind power, wind turbines based on voltage source converters (VSC) have become a typical scenario for grid connection via long-distance AC submarine cables. Due to the extremely wide control bandwidth of power electronic equipment, its interaction with the grid impedance is very likely to induce frequency-range oscillations, ranging from subsynchronous oscillations of a few hertz to high-frequency resonances of several kilohertz.

[0003] In existing technologies, impedance analysis, traditional energy methods, improved energy methods based on filters, and Hilbert-Huang transform (HHT) methods are generally used for oscillation source localization. However, these existing localization methods generally require specific model parameters for localization and have the following drawbacks in practical applications: (1) Impedance analysis method: Since it relies heavily on the detailed control parameters (black box model) provided by the wind turbine manufacturer during use, the parameters are often unavailable or inaccurate in actual engineering, and the offline model cannot reflect the real-time operating conditions. (2) Traditional energy method: In actual engineering, the total active power is calculated based on the power frequency phasor system. However, offshore wind farms often have the phenomenon of "multi-mode coexistence" (such as the simultaneous existence of 20Hz and 500Hz oscillations). Therefore, this traditional method will mix all frequency components during use, causing the energy flow of different frequencies to cancel each other out, resulting in misjudgment. (3) Improved energy method based on filter: This method improves some technologies based on conventional technology by using a bandpass filter to extract specific frequencies. However, conventional IIR / FIR filters introduce significant phase lag during transient processes, resulting in incorrect instantaneous power direction calculations. Furthermore, the filter parameters are fixed and cannot adapt to the frequency drift phenomenon common in offshore wind farms. (4) Hilbert-Huang Transform (HHT) method: Although it has adaptability, it suffers from severe mode aliasing and endpoint effects in strong noise environments, which leads to distortion of decomposition results and makes it impossible to accurately extract the energy characteristics of a single oscillation mode.

[0004] Therefore, this application proposes a broadband oscillation source location method, system, and equipment for offshore wind farms to solve the above-mentioned technical problems. Summary of the Invention

[0005] The main objective of this invention is to provide a method, system, and device for locating broadband oscillation sources in offshore wind farms based on spectral peak clustering VMD and energy flow trend analysis. This method requires no model parameters, can adaptively decouple multimodal data, and is robust to noise and phase hysteresis, thereby solving the technical problems of low positioning accuracy and difficult parameter tuning in multimodal coupling and strong noise environments mentioned in the background art.

[0006] The present invention solves the above-mentioned technical problems by adopting the following technical solutions: A method for locating broadband oscillation sources in offshore wind farms involves the following steps performed using computer equipment: Step S1. Synchronously acquire the three-phase voltage signals of the offshore wind farm's point of common coupling and each feeder branch using a synchronous phasor measurement unit (PMU). The three-phase voltage signals include the time series of the three-phase voltages. and current time series And perform data preprocessing to remove bad data and outliers; Step S2. Perform spectrum analysis on the acquired voltage signal, extract the spectral peaks above the noise floor using Fast Fourier Transform (FFT), and cluster and merge neighboring peaks by setting a frequency tolerance window. Determine the frequency of the cluster center as the oscillation center frequency, and adaptively determine the number of modes for Variational Mode Decomposition (VMD) based on the clustering results (number of clusters). And the initial center frequency of each mode. This step realizes the adaptive initialization of VMD parameters without the need for manual experience setting; Step S3. Based on the determined number of modes Based on the initial center frequency, the original voltage and current signals are decomposed into their initial values ​​using the Variational Mode Decomposition (VMD) algorithm. The intrinsic mode components with finite bandwidth are denoted as voltage mode components. and current mode components ; Step S4. For each modal component, calculate its cumulative interaction energy flow sequence, select a time window segment of the sequence, and perform linear regression using the least squares method to calculate the slope of the energy flow trend. ; Step S5. According to The sign of the node determines the source load property in that mode. When it is determined to be an oscillation source, or when The load was identified as a dissipative load, and the oscillation contribution was calculated based on the proportion of the slope amplitude of multiple nodes, thus finally locating the dominant oscillation source.

[0007] Preferably, in step S2, the peak values ​​are clustered and merged, and the number of decomposition modes is adaptively determined. The specific operating procedures include: The acquired signal undergoes a Fast Fourier Transform (FFT), and the FFT results are then subjected to amplitude scaling and correction. Finally, the amplitude spectrum of the voltage signal is calculated. It is used to identify all local maxima in the spectrum and uses the average amplitude of the entire frequency band as the noise basis. ; Set peak extraction threshold ,in The signal-to-noise ratio coefficient; Extract all amplitudes greater than The local maxima are used as the candidate peak set. Each candidate peak contains a frequency value. and amplitude ; Setting frequency tolerance window For sets Mid-frequency spacing Multiple candidate peaks are grouped into a mode cluster, and the frequency corresponding to the point with the largest amplitude within the cluster is taken as the center frequency of the mode cluster. The final number of modal clusters is denoted as... Set the number of VMD decomposition modes One additional mode is used to absorb background noise or DC components. Preferably, the constrained variational model of the variational mode decomposition algorithm in step S3 is constructed as follows:

[0008]

[0009] in, The set of modal components obtained from the decomposition, The first decomposition obtained Individual and Time Related modal components, The set of center frequencies for each mode. The first decomposition obtained Individual and Time The relevant center frequency, For the Dirac function, This represents the convolution operation. The imaginary unit, This represents the original input signal, i.e., the target real-valued time series data to be decomposed. In the constraint condition (st on the left), it means that the sum of all decomposed modal components should be losslessly restored to the original signal. Regarding time The partial derivative operator is applied to the analytic signal after frequency mixing to perform gradient calculation; This model introduces a quadratic penalty factor and Lagrange multipliers, and iteratively solves the problem using the alternating direction multiplier method (ADMM) under a constrained variational model. Ultimately, it updates the center frequency and bandwidth of each mode, decomposing the original voltage and current signals into a series of intrinsic mode functions (IMFs) with different center frequencies. Then, it selects the voltage mode component containing the dominant oscillation frequency. and current mode components As a target signal, it can effectively filter out broadband background noise.

[0010] Preferably, the slope of the energy flow trend in step S4 The specific calculation process is as follows: First calculate the... Discrete cumulative energy sequence of each mode The calculation formula is as follows:

[0011] in The sampling point number, The sampling interval is... For the first Modal sampling points The modal component sequence after decomposition of the voltage signal. For the first Modal sampling points The modal component sequence of the current signal after decomposition at the location; After selecting the sequence Using the data as the steady-state assessment interval, a linear regression model is constructed as follows: ,in, The intercept of the linear regression model. Represented as the first Discrete cumulative energy of each mode; Solving for the slope using the least squares method The solution formula is as follows:

[0012] in, The number of data points in the steady-state evaluation interval. For data points The corresponding time, For data points The corresponding cumulative energy value.

[0013] Preferably, the energy flow trend linear regression and source-load determination method in step S5 includes: constructing a cumulative interaction energy curve during the calculation of the cumulative interaction energy flow sequence. To perform linear regression fitting; The cumulative interactive energy flow (TEF) curve is calculated based on the extracted single-frequency modal components during the calculation process. The calculation formula is as follows: This curve reflects the direction and magnitude of the cumulative transmission of oscillatory energy within the observation time window; Calculate the slope of the fitted line for linear regression. and the coefficient of determination for linear regression , here Used as a goodness-of-fit criterion coefficient, and a slope threshold is set. and goodness-of-fit threshold ; The confidence level verification operation is performed as follows: like Less than the set threshold If the value is 0.8, then the mode is determined to be a spurious mode or noise and will not be localized. Only when and When this mode is determined to be an effective oscillation mode, it is determined based on the slope. The source load is determined by the positive or negative sign of the signal, and the responsibility is assigned. Otherwise, the mode is determined to be a non-dominant mode or random noise interference, and no source load is assigned.

[0014] Preferably, the slope in step S5 The specific criteria for determining the positive and negative values ​​of source loads are as follows: like Meets the threshold requirement (greater than the set threshold) ), then according to the slope The positive and negative values ​​of the source load are determined as follows: If during the judgment process If the slope is negative, then the device is determined to be emitting oscillating energy into the system, and is thus an oscillation source; If during the judgment process If the slope is positive, then the device is determined to consume oscillating energy and is a damping load.

[0015] Preferably, the formula for calculating the oscillation contribution in step S5 is:

[0016] in, To be identified as an oscillation source (i.e. Normalized contribution of all nodes in ) The set of all nodes identified as oscillation sources. for The node in the node Energy flow trend slope of each node ; Final judgment The largest node is the dominant oscillation source.

[0017] The present invention also provides a broadband oscillation source localization system for offshore wind farms, for performing the steps of any of the methods described above, including: Data acquisition module: used to synchronously acquire three-phase voltage and current signal data of the wind farm common connection point and each feeder branch measured by PMU; Parameter identification module: Used to perform spectral peak clustering and output the number of modes required for VMD decomposition. With the initial center frequency; Mode decomposition module: used to execute the VMD algorithm and output the decoupled single-frequency voltage and current mode components; The responsibility determination module is used to calculate the cumulative energy and perform linear regression analysis, and output the oscillation source location results. Specifically, it includes a trend analysis unit for calculating the energy flow trend slope and assessing confidence, and a source-load discrimination unit for determining responsibility and calculating contribution.

[0018] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0019] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0020] As can be seen from the above technical solution, the present invention provides a method, system, and device for locating broadband oscillation sources in offshore wind farms. Compared with the prior art, the present invention has the following advantages: 1. This invention uses a spectral peak clustering algorithm to adaptively determine VMD parameters, which not only solves the problem of blindly setting the K value manually in the VMD algorithm, but also automatically tracks when the oscillation frequency drifts, ensuring decomposition accuracy, achieving adaptive decoupling, and avoiding mode aliasing.

[0021] 2. This invention introduces the linear regression slope of the accumulated energy flow as a criterion and utilizes the inherent low-pass filtering characteristics of integral operation to cancel out zero-mean random noise during the accumulation process. At this time, the linear regression slope, as a statistical feature, can extract a stable energy transmission trend from the noise-inundated signal, thus enabling the overall positioning process to have strong anti-noise interference capability and robustness.

[0022] 3. This invention employs the VMD algorithm based on non-recursive variational solution. Compared with traditional causal filters, the extracted modal components have no phase lag, which can guarantee the authenticity of the voltage and current phase difference, thereby further ensuring the physical correctness of the energy flow direction calculation.

[0023] 4. By introducing the coefficient of determination of linear regression as the goodness-of-fit index, this invention can automatically eliminate spurious modes with poor fit in the localization process, avoid false alarms of noise modes, and further improve the reliability of the overall system oscillation source localization results.

[0024] 5. In summary, by replacing the traditional instantaneous value judgment with the trend slope, this invention can effectively eliminate random noise interference and the effect of filter phase lag, thereby achieving accurate positioning of the oscillation source under complex working conditions.

[0025] It should be understood that the descriptions in this section are not intended to identify key or essential features of embodiments of the invention, nor are they intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Of course, implementing any product of the invention does not necessarily require achieving all of the advantages described above simultaneously. Attached Figure Description

[0026] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the overall operation flow of the method of the present invention; Figure 2 This is a schematic diagram illustrating the effects of adaptive spectral peak clustering and VMD mode decomposition of the present invention; Figure 3 This is a schematic diagram illustrating the source-load discrimination of the energy flow trend slope based on least squares fitting according to the present invention, used to show the significant difference between the oscillation source and the damping load; Figure 4 This is a comparison diagram of the method of the present invention and the traditional bandpass filter energy method in a strong noise environment; Figure 5 This is a schematic block diagram of the overall structure of the system of the present invention. Detailed Implementation

[0027] 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 a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. 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.

[0028] For details in the embodiments, please refer to Figures 1 to 5 .

[0029] On the one hand, such as Figure 1 As shown, the broadband oscillation source localization method for offshore wind farms proposed in this embodiment of the invention is executed by computer equipment and includes the following steps: Step S1. Broadband Data Acquisition: Synchronously acquire three-phase voltage signals from the offshore wind farm's point of common coupling and each feeder branch using a synchronous phasor measurement unit (PMU). The three-phase voltage signals include the time series of the three-phase voltages. and current time series And perform data preprocessing to remove bad data and outliers.

[0030] In practice, three-phase voltage and current data are collected by broadband measurement units (PMUs) installed at the grid connection point (PCC) and each collector line. The sampling frequency is set to 2.5kHz and the data window length is 2 seconds.

[0031] Step S2. Adaptive Parameter Initialization: Perform spectrum analysis on the acquired voltage signal, extract the spectral peaks above the noise floor using Fast Fourier Transform (FFT), and cluster and merge neighboring peaks by setting a frequency tolerance window. Determine the frequency of the cluster center as the oscillation center frequency, and adaptively determine the number of modes for Variational Mode Decomposition (VMD) based on the clustering results (number of clusters). This step, which sets the initial center frequency of each mode, enables adaptive initialization of VMD parameters without requiring manual experience-based settings.

[0032] The peak values ​​are clustered and merged, and the number of decomposition modes is adaptively determined. The specific operating procedures include: The acquired signal undergoes a Fast Fourier Transform (FFT), and the FFT results are then subjected to amplitude scaling and correction. Finally, the amplitude spectrum of the voltage signal is calculated. It is used to identify all local maxima in the spectrum and uses the average amplitude of the entire frequency band as the noise basis. ; Set peak extraction threshold ,in The signal-to-noise ratio coefficient; Extract all amplitudes greater than The local maxima are used as the candidate peak set. Each candidate peak contains a frequency value. and amplitude ; Setting frequency tolerance window For sets Mid-frequency spacing Multiple candidate peaks are grouped into a mode cluster, and the frequency corresponding to the point with the largest amplitude within the cluster is taken as the center frequency of the mode cluster. The final number of modal clusters is denoted as... Set the number of VMD decomposition modes One additional mode is used to absorb background noise or DC components.

[0033] At this point, the effects of adaptive spectral peak clustering and VMD mode decomposition are as follows: Figure 2 As shown, in the actual implementation process, the collected phase A voltage data... A Fast Fourier Transform (FFT) analysis is performed to obtain its spectral characteristics. In this embodiment, the specific execution process is as follows: (1) Calculate the noise floor and set threshold: First, calculate the average amplitude across the entire frequency band as the noise floor. In the measured data of this embodiment, the following calculations were obtained: ; Then set the signal-to-noise ratio coefficient. .suggestion The value range is from 1.2 to 1.5. Too small a value can introduce noise peaks, while too large a value can easily miss weak signals. This embodiment uses... (Conservative setting for environments with strong interference), then the peak extraction threshold is:

[0034] (2) Peak search and clustering: Search the spectrum for all amplitudes greater than Local maxima were identified, and a set of significant peak frequencies was preliminarily identified. .

[0035] Setting frequency tolerance window .suggestion The value range is 0.5Hz to 2.0Hz, used to handle the "single-peak splitting" phenomenon caused by spectral leakage or frequency drift. This embodiment sets... .

[0036] For sets Perform a traversal check: for and Calculate the frequency difference between the two peaks. .

[0037] because (2Hz) It is determined that the two peaks belong to the same oscillation mode (i.e., spectral splitting has occurred), and they are grouped into the same "subsynchronous oscillation mode cluster".

[0038] Within this cluster, select the frequency corresponding to the point with the largest amplitude (in this example, ). () is used as the center frequency of the cluster.

[0039] The peak value is far from other peak values, and can be considered as a separate "high-frequency oscillation mode cluster".

[0040] (3) Determine VMD parameters: The final number of modal clusters was counted. (Corresponding to 24Hz and 580Hz respectively).

[0041] Set the number of VMD decomposition modes One additional mode is used to absorb background noise or DC components.

[0042] Finally, the initial center frequency vector of the output VMD algorithm is: .

[0043] By using the spectral peak clustering algorithm to adaptively determine the VMD parameters, not only is the blindness of manually setting the K value in the VMD algorithm resolved, but it can also automatically track when the oscillation frequency drifts (such as from 23Hz to 25Hz), ensuring decomposition accuracy, achieving adaptive decoupling, and avoiding mode aliasing.

[0044] Step S3. Orthogonal signal decoupling: based on a determined number of modes Based on the initial center frequency, the original voltage and current signals are decomposed into their initial values ​​using the Variational Mode Decomposition (VMD) algorithm. The intrinsic mode components with finite bandwidth are denoted as voltage mode components. and current mode components .

[0045] The constrained variational model of the variational mode decomposition algorithm is constructed as follows:

[0046]

[0047] in, The set of modal components obtained from the decomposition, The first decomposition obtained Individual and Time Related modal components, The set of center frequencies for each mode. The first decomposition obtained Individual and Time The relevant center frequency, For the Dirac function, This represents the convolution operation. The imaginary unit, This represents the original input signal, i.e., the target real-valued time series data to be decomposed. In the constraint condition (st on the left), it means that the sum of all decomposed modal components should be losslessly restored to the original signal. Regarding time The partial derivative operator is applied to the analytic signal after frequency mixing to perform gradient calculation.

[0048] This model introduces a quadratic penalty factor and Lagrange multipliers, and iteratively solves the problem using the alternating direction multiplier method (ADMM) under a constrained variational model. Ultimately, it updates the center frequency and bandwidth of each mode, decomposing the original voltage and current signals into a series of intrinsic mode functions (IMFs) with different center frequencies. Then, it selects the voltage mode component containing the dominant oscillation frequency. and current mode components As a target signal, it can effectively filter out broadband background noise.

[0049] In practical implementation, the VMD algorithm is used to decompose the signal, resulting in IMF1 (center frequency converges to 24.1Hz) and IMF2 (center frequency converges to 581Hz). During the decomposition process, a quadratic penalty factor is applied. Set it to 2000 to ensure better frequency band separation.

[0050] Step S4. Trend Feature Extraction: For each modal component, calculate its cumulative interaction energy flow sequence, select a time window segment of the sequence, and use the least squares method to perform linear regression to calculate the slope of the energy flow trend. .

[0051] Among them, the slope of the energy flow trend The specific calculation process is as follows: First calculate the... Discrete cumulative energy sequence of each mode The calculation formula is as follows:

[0052] in The sampling point number, The sampling interval is... For the first Modal sampling points The modal component sequence after decomposition of the voltage signal. For the first Modal sampling points The modal component sequence of the current signal after decomposition at the location; After selecting the sequence Using the data as the steady-state assessment interval, a linear regression model is constructed as follows: ,in, The intercept of the linear regression model. Represented as the first Discrete cumulative energy of each mode; Solving for the slope using the least squares method The solution formula is as follows:

[0053] in, The number of data points in the steady-state evaluation interval. For data points The corresponding time, For data points The corresponding cumulative energy value.

[0054] At this point, the cumulative interaction energy curve can be constructed during the calculation of the cumulative interaction energy flow sequence. To perform linear regression fitting; The cumulative interactive energy flow (TEF) curve is calculated based on the extracted single-frequency modal components during the calculation process. The calculation formula is as follows: The curve reflects the direction and magnitude of the cumulative transmission of oscillating energy within the observation time window.

[0055] Specifically, in the actual implementation process, the analysis is conducted for the IMF1 (24Hz) subsynchronous mode: (1) Calculate the cumulative interaction energy flow sequence .

[0056] (2) The latter part of the data (1.0s to 2.0s) is taken as the steady-state interval.

[0057] (3) Fitting the line using the least squares method: Fan A branch: Fitting slope There is a coefficient of determination. .

[0058] Fan B branch: Fitted slope There is a coefficient of determination. .

[0059] Step S5. Determination of Responsibility and Output of Results: Based on The sign of the node determines the source load property in that mode. When it is determined to be an oscillation source, or when The load was identified as a dissipative load, and the oscillation contribution was calculated based on the proportion of the slope amplitude of multiple nodes, thus finally locating the dominant oscillation source.

[0060] The linear regression method for energy flow trends and source load determination includes: Based on the slope of the fitted line of the linear regression and the coefficient of determination for linear regression , here Used as a goodness-of-fit criterion, setting a slope threshold. and goodness-of-fit threshold ; The confidence level verification operation is performed as follows: like Less than the set threshold If the value is 0.8, then the mode is determined to be a spurious mode or noise and will not be localized. Only when and When this mode is determined to be an effective oscillation mode, it is determined based on the slope. The source load is determined by the sign of the positive or negative value, and the responsibility is assigned. Otherwise, the mode is determined to be a non-dominant mode or random noise interference, and no source load assignment is performed. In this case, the slope is... The specific criteria for determining the positive and negative values ​​of source loads are as follows: like Meets the threshold requirement (greater than the set threshold) ), then according to the slope The positive and negative values ​​of the source load are determined as follows: If during the judgment process If the slope is negative, then the device is determined to be emitting oscillating energy into the system, and is thus an oscillation source; If during the judgment process If the slope is positive, then the device is determined to consume oscillating energy and is a damping load.

[0061] At this point, the significant difference between the oscillation source and the damping load is as follows: Figure 3 As shown.

[0062] By introducing the coefficient of determination of linear regression as the goodness-of-fit index, spurious modes with poor fit can be automatically eliminated in the localization process, avoiding false alarms of noise modes and further improving the reliability of the overall system oscillation source localization results.

[0063] At this point, based on the identified oscillation source, the oscillation contribution is further calculated, and the final formula for calculating the oscillation contribution is:

[0064] in, To be identified as an oscillation source (i.e. Normalized contribution of all nodes in ) The set of all nodes identified as oscillation sources. for The node in the node Energy flow trend slope of each node ; Final judgment at this point The largest node is the dominant oscillation source.

[0065] In actual implementation, the energy flow trend slope calculated in step S4 is used as a basis. and linear regression determination coefficient The source and load properties of each monitoring node are determined.

[0066] (1) Set the judgment threshold: Set slope threshold and goodness-of-fit threshold .

[0067] Goodness-of-fit threshold It is recommended to set it to 0.8 to 0.9. When When this occurs, it indicates that the cumulative energy flow curve exhibits a significant linear monotonic trend, eliminating the interference of random fluctuations. This embodiment sets... .

[0068] Slope threshold Used to filter out non-dominant modes with extremely weak energy. This embodiment is set as follows: .

[0069] (2) Execute the decision logic (taking the 24Hz mode as an example): For fan A (monitoring point 1): The slope in this mode was calculated. Coefficient of determination .

[0070] Judgment process: First, check the confidence level, because and Determine the mode This is a valid oscillation mode. Next, check the sign of the slope, since... This indicates that energy is continuously flowing out, and it is determined that fan A is the oscillation source (exhibiting negative damping characteristics and injecting energy into the system).

[0071] For fan B (monitoring point 2): The slope in this mode was calculated. Coefficient of determination .

[0072] Judgment process: It also meets the validity threshold. Because... This indicates that energy is continuously flowing in and being consumed, and wind turbine B is determined to be a damping load (consuming oscillating energy).

[0073] Based on the above findings, the system outputs a location report: Wind turbine A is the dominant oscillation source causing this 24Hz subsynchronous oscillation. According to this conclusion, maintenance personnel can specifically perform turbine tripping operations on wind turbine A or adjust its converter control parameters (such as reducing the proportional gain). This suppresses oscillations.

[0074] Furthermore, it is worth noting that existing technologies mostly determine the source load based on the direction of instantaneous power. Under strong background noise, the instantaneous power waveform is often chaotic, causing repeated jumps in positive and negative signs, which can easily lead to misjudgment. This method introduces the linear regression slope of the accumulated energy flow as a criterion. By utilizing the inherent low-pass filtering characteristics of integral operations, it can cancel out zero-mean random noise during the accumulation process. At this time, the linear regression slope, as a statistical feature, can extract a stable energy transmission trend from the noise-inundated signal. Therefore, it can make the overall positioning process have strong anti-noise interference capability and robustness. Moreover, experiments show that even under harsh conditions with a signal-to-noise ratio (SNR) as low as 5dB, this method can still maintain a liability determination accuracy of over 95%, which is significantly better than traditional methods.

[0075] Based on the above operations, combined with Figure 4 If the same data is processed using a traditional bandpass filter, the filter introduces a phase lag of approximately 30 degrees near 24Hz, causing the instantaneous power curve of wind turbine A to fluctuate wildly around the 0 axis. The integrated energy curve exhibits an oscillating divergence, making linear regression impossible and difficult to determine the source of the oscillation. In contrast, this method, through VMD phase-lag-free decomposition combined with slope statistics, clearly identifies the oscillation source, verifying the superiority of this method. Furthermore, by employing the VMD algorithm based on non-recursive variational solution, compared to traditional causal filters, the extracted modal components have no phase lag, ensuring the authenticity of the voltage and current phase differences, thereby further ensuring the physical correctness of the energy flow direction calculation.

[0076] In summary, this method, by replacing the traditional instantaneous value judgment with the trend slope, can effectively eliminate random noise interference and the effect of filter phase lag, thereby achieving accurate positioning of the oscillation source under complex working conditions.

[0077] On the other hand, such as Figure 5 As shown, the present invention also discloses a broadband oscillation source localization system for offshore wind farms, used to perform the steps of any of the methods described above, including: Data acquisition module: used to synchronously acquire three-phase voltage and current signal data of the wind farm common connection point and each feeder branch measured by PMU; Parameter identification module: Used to perform spectral peak clustering and output the number of modes required for VMD decomposition. With the initial center frequency; Mode decomposition module: used to execute the VMD algorithm and output the decoupled single-frequency voltage and current mode components; The responsibility determination module is used to calculate the cumulative energy and perform linear regression analysis, and output the oscillation source location results. Specifically, it includes a trend analysis unit for calculating the energy flow trend slope and assessing confidence, and a source-load discrimination unit for determining responsibility and calculating contribution.

[0078] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0079] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0080] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the broadband oscillation source localization methods for offshore wind farms described above.

[0081] It is understood that the system provided in the embodiments of the present invention corresponds to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant content can be referred to the corresponding parts of the above method.

[0082] This application also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, communication interface, and memory communicate with each other via the communication bus. Memory, used to store computer programs; The processor, when executing the program stored in the memory, implements the above-mentioned method for locating broadband oscillation sources in offshore wind farms.

[0083] The communication bus mentioned in the above-mentioned electronic devices can be a standard bus for interconnecting peripheral components or an extended industrial standard structure bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.

[0084] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0085] The memory may include random access memory or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0086] The processors mentioned above can be general-purpose processors, including central processing units, network processors, etc.; they can also be digital signal processors, application-specific integrated circuits, field-programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0087] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0088] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0089] Furthermore, it should be noted that if any directional indication (such as up, down, left, right, front, back, etc.) is involved in the embodiments of the present invention, the directional indication is only used to explain the relative positional relationship and movement of each component in a specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0090] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, in the embodiments of this invention, "multiple" refers to two or more. Moreover, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

Claims

1. A method for locating broadband oscillation sources in offshore wind farms, characterized in that, include: The three-phase voltage signals of the wind farm's common coupling point and each feeder branch are acquired synchronously. The spectral peaks above the noise floor are extracted and clustered to adaptively determine the number of modes in the variational mode decomposition. and the initial center frequency of each mode; Based on mode number And the initial center frequency, the original voltage signal and the original current signal are decomposed into their respective values ​​using the variational mode decomposition algorithm. A voltage mode component and a current mode component with finite bandwidth; For each modal component, its cumulative interaction energy flow sequence is calculated. A time window segment of this sequence is selected for linear regression to calculate the slope of the energy flow trend. ; according to The positive or negative sign of the value determines whether the source load of the node in this mode is an oscillation source or a dissipative load, and the oscillation contribution is calculated based on the proportion of the slope amplitude of multiple nodes, and finally the dominant oscillation source is located.

2. The method for locating broadband oscillation sources in offshore wind farms as described in claim 1, characterized in that, The peak values ​​are clustered and merged, and the number of decomposition modes is adaptively determined. The specific operating procedures include: Calculate the amplitude spectrum of the voltage signal The average amplitude across the entire frequency band is used as the noise floor. ; Set peak extraction threshold , The signal-to-noise ratio coefficient; Extract all amplitudes greater than The local maxima are used as the candidate peak set. Each candidate peak contains a frequency value. and amplitude ; Setting frequency tolerance window For sets Mid-frequency spacing Multiple candidate peaks are grouped into a mode cluster, and the frequency corresponding to the point with the largest amplitude within the cluster is taken as the center frequency of the mode cluster. The final number of modal clusters is denoted as... Set the number of VMD decomposition modes The additional modes are used to absorb background noise or DC components.

3. The method for locating broadband oscillation sources in offshore wind farms as described in claim 2, characterized in that, The constrained variational model of the variational mode decomposition algorithm is constructed as follows: in, The set of modal components obtained from the decomposition, The first decomposition obtained Individual and Time Related modal components, The set of center frequencies for each mode. The first decomposition obtained Individual and Time The relevant center frequency, For the Dirac function, This represents the convolution operation. The imaginary unit, This is represented as the original input signal, i.e., the target real-valued time series data to be decomposed. The constraints state that the sum of all decomposed modal components should be losslessly restored to the original signal. Regarding time The partial derivative operator is applied to the analytic signal after frequency mixing to perform gradient calculation; The model is solved iteratively using the alternating direction multiplier method.

4. The method for locating broadband oscillation sources in offshore wind farms as described in claim 3, characterized in that, The slope of the energy flow trend The specific calculation process is as follows: First calculate the... Discrete cumulative energy sequence of each mode The calculation formula is as follows: in The sampling point number, The sampling interval is... For the first Modal sampling points The modal component sequence after decomposition of the voltage signal. For the first Modal sampling points The modal component sequence of the current signal after decomposition at the location; After selecting the sequence Using the data as the steady-state assessment interval, a linear regression model is constructed as follows: ,in, The intercept of the linear regression model. Represented as the first Discrete cumulative energy of each mode; Solving for the slope using the least squares method The solution formula is as follows: in, The number of data points in the steady-state evaluation interval. For data points The corresponding time, For data points The corresponding cumulative energy value.

5. The method for locating broadband oscillation sources in offshore wind farms as described in claim 3, characterized in that, The energy flow trend linear regression and source-load determination method includes: constructing a cumulative interaction energy curve during the calculation of the cumulative interaction energy flow sequence. For linear regression fitting, the calculation formula is as follows: ; Calculate the slope of the fitted line and the coefficient of determination for linear regression And set a slope threshold. and goodness-of-fit threshold ; The confidence level verification operation is performed as follows: like Less than the set threshold If the mode is not identified as a false mode or noise, it will not be localized. Only when and When this mode is determined to be an effective oscillation mode, it is determined based on the slope. The positive or negative value of the load is determined and the responsibility assignment result is output. Otherwise, the mode is determined to be a non-dominant mode or random noise interference, and no source load assignment is performed.

6. The method for locating broadband oscillation sources in offshore wind farms as described in claim 5, characterized in that, The slope The specific criteria for determining the positive or negative sign are as follows: If during the judgment process If the slope is negative, then the device is determined to be emitting oscillating energy into the system, and is thus an oscillation source; If during the judgment process If the slope is positive, then the device is determined to consume oscillating energy and is a damping load.

7. The method for locating broadband oscillation sources in offshore wind farms as described in claim 6, characterized in that, The formula for calculating the oscillation contribution is as follows: in, The normalized contribution of all nodes identified as oscillation sources. The set of all nodes identified as oscillation sources. for The node in the node Energy flow trend slope of each node ; Final judgment The largest node is the dominant oscillation source.

8. A broadband oscillation source localization system for offshore wind farms, used to perform the steps of the method as described in any one of claims 1 to 7, characterized in that, include: Data acquisition module: used to synchronously acquire three-phase voltage and current signal data of the wind farm common connection point and each feeder branch measured by PMU; Parameter identification module: Used to perform spectral peak clustering and output the number of modes required for VMD decomposition. With the initial center frequency; Mode decomposition module: used to execute the VMD algorithm and output the decoupled single-frequency voltage and current mode components; The liability determination module is used to calculate the cumulative energy and perform linear regression analysis, and output the oscillation source location results.

9. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 7.

Images