Method and device for identifying sub / super-synchronous oscillation signals of power system

Abstract

The application discloses a power system sub / super-synchronous oscillation signal identification method and device, and aims to solve the technical problem that current related technologies cannot effectively monitor and accurately distinguish sub-synchronous and super-synchronous oscillation signals. Multi-source time sequence information of a power system is collected, and a fusion representation vector is constructed according to the multi-source time sequence information; the multi-source time sequence information is subjected to frequency spectrum analysis and enhancement, and an oscillation signal containing a sub / super-synchronous frequency band is obtained; based on the frequency spectrum analysis result in the frequency spectrum analysis and enhancement process, the oscillation mode parameters of the oscillation signal in the sub / super-synchronous frequency band are identified, and the oscillation risk level is evaluated based on the oscillation mode parameters; the fusion representation vector is fused and verified according to the multi-source time sequence information and the oscillation mode parameters, the sub / super-synchronous oscillation region and the time sequence evolution path are identified; the oscillation mode parameters are subjected to reverse verification, and when the reverse verification is passed, various types of identification information obtained in the previous steps are output as the multi-dimensional identification result of the oscillation signal.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system signal processing, and in particular to a power system sub / super-synchronous oscillation signal identification method, a power system sub / super-synchronous oscillation signal identification device, an electronic device and a storage medium. BACKGROUND

[0002] With the continuous expansion of the scale of the power system and the continuous improvement of the intelligence level of power equipment, the stability of the power system is facing increasingly severe challenges. Especially the dynamic phenomena such as sub-synchronous and super-synchronous oscillation. These oscillations not only cause the power system to operate inefficiently, but also may cause safety accidents such as damage and shutdown of power equipment. Therefore, accurately identifying and monitoring the oscillation signals in the power system, especially the sub-synchronous and super-synchronous frequency band oscillation signals, is crucial for the safe and stable operation of the power system.

[0003] At present, the oscillation signals in the power system are usually collected by devices such as synchronous phasor measurement units (PMU) and wave recorders. The synchronous phasor measurement unit can monitor the key parameters such as voltage and current in the power grid in real time. However, it can only capture low-frequency synchronous oscillation signals and cannot effectively monitor sub-synchronous (frequency lower than system fundamental frequency) and super-synchronous (frequency higher than system fundamental frequency) oscillation signals. Although the wave recorder can obtain high-frequency oscillation signals, its monitoring range is limited by factors such as sampling frequency and device accuracy.

[0004] In the extraction and identification of oscillation signals, current methods mostly rely on traditional frequency spectrum analysis techniques. Such as Fast Fourier Transform (FFT). However, these methods often have difficulty in accurately distinguishing sub-synchronous and super-synchronous oscillations when faced with spectral overlap or complex signals. SUMMARY

[0005] The present application provides a power system sub / super-synchronous oscillation signal identification method, a power system sub / super-synchronous oscillation signal identification device, an electronic device and a storage medium, which are used to solve or partially solve the technical problem that current related technologies cannot effectively monitor and accurately distinguish sub-synchronous and super-synchronous oscillation signals.

[0006] The present application provides a power system sub / super-synchronous oscillation signal identification method, which comprises:

[0007] Collecting multi-source time series information of the power system, and constructing a fusion representation vector according to the multi-source time series information;

[0008] Performing frequency spectrum analysis and enhancement on the multi-source time series information to obtain oscillation signals containing sub / super-synchronous frequency bands;

[0009] Based on the spectrum analysis result obtained in the spectrum analysis and enhancement process, identify the oscillation mode parameter of the oscillation signal in the sub / super synchronous frequency band, and evaluate the oscillation risk level based on the oscillation mode parameter;

[0010] According to the multi-source time sequence information and the oscillation mode parameter, the fusion representation vector is fused and verified, and the sub / super synchronous oscillation area and time sequence evolution path are identified;

[0011] The oscillation mode parameter is reverse checked, and when the reverse check is passed, the oscillation mode parameter, the oscillation risk level, the sub / super synchronous oscillation area and the time sequence evolution path are output as the multi-dimensional identification result of the oscillation signal.

[0012] Optionally, the multi-source time sequence information includes recorded wave signal, synchronous phasor signal, protection event action signal and device state quantity; the fusion representation vector is constructed according to the multi-source time sequence information, including:

[0013] The time stamps of the synchronous phasor signal are taken as reference axes, and the recorded wave signal, the synchronous phasor signal, the protection event action signal and the device state quantity are time aligned and time reconstructed to construct a plurality of multi-source data corresponding to different time periods;

[0014] For the multi-source data in each time period, the time sequence characteristic variable of the multi-source data is extracted, and the time sequence characteristic variable is mapped to construct a fusion representation vector.

[0015] Optionally, the multi-source time sequence information includes recorded wave signal; the multi-source time sequence information is subjected to spectrum analysis and enhancement to obtain the oscillation signal containing the sub / super synchronous frequency band, including:

[0016] The recorded wave signal is filtered by a pre-constructed band-pass filter to obtain a sub / super synchronous frequency band;

[0017] The recorded wave signal is subjected to center frequency shift processing, and the center frequency of the sub / super synchronous frequency band is mapped to a zero frequency point to obtain a frequency-shifted recorded wave signal;

[0018] The frequency-shifted recorded wave signal is resampled, and the resampled signal is subjected to spectrum analysis by short-time Fourier transform;

[0019] The frequency-shifted recorded wave signal is subjected to spectrum enhancement according to the spectrum analysis result to obtain the oscillation signal in the sub / super synchronous frequency band.

[0020] Optionally, the oscillation risk level is evaluated based on the oscillation modal parameter of the oscillation signal in the sub- / ultra-synchronous frequency band, comprising:

[0021] A multi-modal signal model is constructed by considering the frequency, amplitude, damping coefficient and initial phase angle of the multi-modal oscillation.

[0022] A loss function is constructed by minimizing the square error between the oscillation signal and the model signal output by the multi-modal signal model.

[0023] The constraint conditions of the loss function are constructed based on the amplitude constraint, frequency constraint and damping ratio constraint.

[0024] The frequency analysis results in the frequency analysis and enhancement process are obtained, the frequency of the frequency analysis results is extracted as the initial frequency, and the frequency, amplitude, initial phase angle and damping ratio of the multi-modal oscillation are obtained by optimizing the loss function under the constraint conditions, as the oscillation modal parameter of the oscillation signal in the sub- / ultra-synchronous frequency band.

[0025] The oscillation risk level of the oscillation signal is evaluated based on the frequency, amplitude, initial phase angle and damping ratio of the multi-modal oscillation.

[0026] Optionally, the multi-source time series information includes a synchronous phasor signal; and the fusion verification of the fusion representation vector based on the multi-source time series information and the oscillation modal parameter to identify the sub- / ultra-synchronous oscillation region and the time evolution path, comprising:

[0027] The time correlation of the oscillation modal parameter and the synchronous phasor signal in different time windows is calculated based on a preset sliding window length.

[0028] The target time window when the time correlation exceeds a preset correlation threshold is identified as the record starting point of the sub- / ultra-synchronous oscillation region, and the identification of the time evolution path is triggered.

[0029] The first derivative of the time correlation is analyzed in real time, and the evolution inflection point of the time evolution path is determined based on the transformation rate analysis result.

[0030] In the time evolution path identification process, the fusion verification analysis of the fusion representation vector based on the principal component analysis is performed based on the oscillation modal parameter and the synchronous phasor signal, to obtain a principal component contribution degree matrix and a principal component time series change matrix.

[0031] The contribution degree analysis of the principal component contribution degree matrix is performed to determine the final sub- / ultra-synchronous oscillation region.

[0032] According to the principal component time sequence change matrix, a trend of change of the oscillation mode over time is tracked, and a final time sequence evolution path is determined according to positions corresponding to principal components at different times.

[0033] Optionally, the principal component analysis-based fusion verification analysis is performed on the fusion representation vector based on the oscillation mode parameter and the synchrophasor signal, a principal component contribution degree matrix and a principal component time sequence change matrix are obtained, and the method comprises the following steps:

[0034] A data matrix is constructed according to a time sequence of the oscillation mode parameter and the synchrophasor signal, and a feature matrix is obtained by performing standardization processing on the data matrix;

[0035] A covariance matrix is constructed according to the feature matrix, and a plurality of eigenvalues and a feature vector corresponding to each eigenvalue are obtained by performing eigenvalue decomposition on the covariance matrix;

[0036] The feature matrix is sorted according to the order of eigenvalues from large to small to construct a principal component contribution degree matrix, and a feature vector space matrix corresponding to the principal component contribution degree matrix is constructed based on each feature vector;

[0037] The fusion representation vector is projected into the feature vector space matrix to obtain a principal component time sequence change matrix.

[0038] Optionally, the multi-source time sequence information comprises a protection event action signal; the oscillation mode parameter is reversely verified, and when the reverse verification passes, the oscillation mode parameter, the oscillation risk level, the sub- / ultra-synchronous oscillation region and the time sequence evolution path are output as a multi-dimensional recognition result of the oscillation signal, and the method comprises the following steps:

[0039] A time backtracking window is constructed based on a trigger time of the protection event action signal, with the trigger time as an anchor point;

[0040] The power spectrum energy density of the oscillation mode parameter in the time backtracking window is calculated;

[0041] When the power spectrum energy density is greater than a preset energy density threshold and the multi-mode oscillation frequency is located in the sub- / ultra-synchronous frequency band, it is determined that the multi-mode trajectory of the oscillation signal is associated with the protection action, and the oscillation mode parameter, the oscillation risk level, the sub- / ultra-synchronous oscillation region and the time sequence evolution path are output as the multi-dimensional recognition result of the oscillation signal.

[0042] The application further provides an electric power system sub- / ultra-synchronous oscillation signal identification device, which comprises:

[0043] The fusion representation vector construction unit is configured to collect multi-source time sequence information of a power system and construct a fusion representation vector according to the multi-source time sequence information.

[0044] The spectrum analysis and enhancement unit is configured to perform spectrum analysis and enhancement on the multi-source time sequence information to obtain an oscillation signal containing a sub- / ultra-synchronous frequency band.

[0045] The oscillation mode parameter identification unit is configured to identify an oscillation mode parameter of the oscillation signal in the sub- / ultra-synchronous frequency band based on a spectrum analysis result obtained in the spectrum analysis and enhancement process, and evaluate an oscillation risk level based on the oscillation mode parameter.

[0046] The fusion verification unit is configured to perform fusion verification on the fusion representation vector according to the multi-source time sequence information and the oscillation mode parameter, identify a sub- / ultra-synchronous oscillation region and a time sequence evolution path.

[0047] The reverse verification unit is configured to perform reverse verification on the oscillation mode parameter, and when the reverse verification passes, output the oscillation mode parameter, the oscillation risk level, the sub- / ultra-synchronous oscillation region and the time sequence evolution path as a multi-dimensional identification result of the oscillation signal.

[0048] The present application also provides an electronic device, the device comprising a processor and a memory:

[0049] The memory is configured to store program code and transmit the program code to the processor.

[0050] The processor is configured to execute the power system sub- / ultra-synchronous oscillation signal identification method according to the instructions in the program code.

[0051] The present application also provides a computer readable storage medium for storing program code, the program code being used to execute the power system sub- / ultra-synchronous oscillation signal identification method.

[0052] From the above technical solutions, the present application has the following advantages:

[0053] A power system sub / super-synchronous oscillation signal identification method is provided. First, multi-source time sequence information of the power system is collected, and a fusion representation vector is constructed according to the multi-source time sequence information. Thus, by organically integrating the monitored multi-class data sources, the identification of the oscillation signal is more comprehensive and robust, and the time alignment and fusion modeling between the data can also be realized, providing a unified basis for subsequent analysis. Then, the multi-source time sequence information is subjected to frequency spectrum analysis and enhancement to obtain an oscillation signal containing a sub / super-synchronous frequency band. Thus, by using the frequency spectrum analysis and enhancement means, not only the oscillation signal containing the sub / super-synchronous frequency band can be obtained, but also the identification sensitivity to weak signals can be significantly improved, which is conducive to discovering potential system instability factors at an early stage. Based on the frequency spectrum analysis results obtained in the frequency spectrum analysis and enhancement process, the oscillation modal parameters of the oscillation signal in the sub / super-synchronous frequency band are identified, and the oscillation risk level is evaluated based on the oscillation modal parameters, so as to evaluate the current risk situation of the system, so that the high-risk situation can be responded to in time. Then, the fusion representation vector is subjected to fusion verification according to the multi-source time sequence information and the oscillation modal parameters, not only the sub / super-synchronous oscillation region and the time sequence evolution path can be identified, but also the accuracy and reliability of the identification result can be improved, and the false alarm and missed alarm risk can be reduced. Finally, the oscillation modal parameters are subjected to reverse verification, and when the reverse verification passes, the oscillation modal parameters, the oscillation risk level, the sub / super-synchronous oscillation region and the time sequence evolution path are output as the multi-dimensional identification result of the oscillation signal, which has practical regulation significance. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, a brief introduction will be given below to the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0055] Figure 1 A flow chart of steps of a power system sub / super-synchronous oscillation signal identification method;

[0056] Figure 2 A schematic diagram of the overall process of a power system sub / super-synchronous oscillation signal identification method;

[0057] Figure 3 A structural block diagram of a power system sub / super-synchronous oscillation signal identification device. DETAILED DESCRIPTION

[0058] The embodiment of the present application provides a power system sub / super synchronous oscillation signal identification method, a power system sub / super synchronous oscillation signal identification device, an electronic device and a storage medium, and is used for solving or partially solving the technical problem that current related technologies cannot effectively monitor and accurately distinguish sub-synchronous and super-synchronous oscillation signals.

[0059] In order to make the application purpose, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the following described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0060] In order to make those skilled in the art better understand the technical solutions provided by the embodiments of the present application, first, some technical features involved in the solutions will be briefly described:

[0061] Sub / super synchronous oscillation region: a concept used to describe the frequency range of oscillation in a power system, corresponding to the oscillation phenomenon below and above the synchronous frequency respectively. These oscillations may affect the stability of the power system. Therefore, special attention and prevention are needed in system design and operation.

[0062] Time sequence evolution path: the time sequence evolution path is a multi-dimensional dynamic process in the identification of power system sub / super synchronous oscillation. It not only contains amplitude changes, but also reflects the evolution law of other oscillation modal characteristics (frequency, phase, damping ratio, etc.) over time.

[0063] As an example, at present, the oscillation signals in the power system are usually collected by devices such as synchronous phasor measurement units PMU and wave recorders. The synchronous phasor measurement unit can monitor the key parameters such as voltage and current in the power grid in real time. However, it can only capture the lower frequency synchronous oscillation signal and cannot effectively monitor the sub-synchronous (frequency lower than the system fundamental frequency) and super-synchronous (frequency higher than the system fundamental frequency) oscillation signals. Although the wave recorder can obtain high-frequency oscillation signals, its monitoring range is limited by factors such as sampling frequency and device accuracy.

[0064] In the extraction and identification of oscillation signals, current methods mostly rely on traditional frequency spectrum analysis techniques. For example, fast Fourier transform FFT. However, these methods often have difficulty in accurately distinguishing sub-synchronous and super-synchronous oscillation when facing spectrum overlap or complex signals.

[0065] Therefore, one of the core points of the embodiments of the present application is that, in view of the deficiencies of the prior art, a power system sub / super-synchronous oscillation signal accurate identification method based on multi-source data fusion is proposed. First, by organically integrating the monitored multi-type data sources, the identification of the oscillation signal is more comprehensive and robust, and the time alignment and fusion modeling between the data can also be achieved, providing a unified basis for subsequent analysis. Then, by using spectrum analysis and enhancement means, not only the oscillation signal containing the sub / super-synchronous frequency band can be obtained, but also the identification sensitivity to weak signals can be significantly improved, which is conducive to discovering potential system instability factors in the early stage. Then, based on the spectrum analysis results obtained in the spectrum analysis and enhancement process, the oscillation modal parameters of the oscillation signal in the sub / super-synchronous frequency band are identified by optimization, and the oscillation risk level is further evaluated based on the oscillation modal parameters, so as to evaluate the current risk situation of the system, so that the high-risk situation can be responded in time. Then, the multi-modal oscillation parameters and the multi-dimensional dynamic characteristics of the synchronous phasor signal are combined to realize fusion verification, identify the sub / super-synchronous oscillation region and time sequence evolution path, which can improve the accuracy and reliability of the identification result and reduce the risk of false alarm and missed alarm. Finally, combined with the protection action, the reverse verification based on the causal correlation analysis is carried out, and when the reverse verification passes, the oscillation modal parameters, the oscillation risk level, the sub / super-synchronous oscillation region and the time sequence evolution path are output as the multi-dimensional identification result of the oscillation signal, which has practical regulation significance.

[0066] Referring to Figure 1 , a step flowchart of a power system sub / super-synchronous oscillation signal identification method provided by the embodiments of the present application is shown, which can specifically include the following steps:

[0067] Step 101, collecting multi-source time sequence information of the power system, and constructing a fusion representation vector according to the multi-source time sequence information;

[0068] In actual application, various types of time sequence data from the wave recorder, the synchronous phasor measurement device, the protection action recording device and the equipment state data measurement device during the operation of the power system are collected.

[0069] Therefore, the multi-source time sequence information proposed in the embodiments of the present application can mainly include the wave recording signal from the wave recorder , the synchronous phasor signal from the synchronous phasor measurement device , the protection event action signal from the protection action recording , and the equipment state quantity from the equipment state measurement device .

[0070] After obtaining multi-source time-series information, time alignment processing can be performed on this data to establish a data mapping relationship with a unified time scale, and a fusion representation vector can be constructed through feature extraction. The constructed fusion representation vector includes key variables such as voltage, current, frequency, phase angle, power, and switching state.

[0071] In some embodiments, the multi-source timing information includes waveform recording signals, synchronization phasor signals, protection event action signals, and device status variables; the implementation process of constructing a fused representation vector based on the multi-source timing information mainly includes the following sub-steps S01 to S02:

[0072] Step S01: Using the timestamp of the synchronization phasor signal as the reference axis, perform time alignment and time reconstruction on the waveform recording signal, synchronization phasor signal, protection event action signal and equipment status quantity to construct multiple multi-source data corresponding to different time periods;

[0073] The timestamp of the synchronization phasor signal can be selected. As a reference timeline, alignment and time reconstruction are achieved for other data sources using methods such as linear interpolation, polynomial interpolation, or sliding window averaging. During time alignment and reconstruction, the synchronization error is less than a preset error threshold (e.g., ...). The uniform resampling interval is set to Based on window length Slide to construct cross-source data blocks for each time period (which can also be understood as time slices), i.e., multi-source data.

[0074] Step S02: For multi-source data in each time period, extract the temporal feature variables of the multi-source data, and perform time mapping based on the temporal feature variables to construct a fusion representation vector.

[0075] Construct the following time mapping function for time mapping to obtain the fused representation vector:

[0076]

[0077] in, Represents time-based mapping function For time series characteristic variables The fused representation vector obtained after time mapping processing; express Align from low sampling rate sources using linear interpolation or moving average. The reconstructed value.

[0078] Therefore, for each multi-source data within the same time period, we can first extract time-series feature variables, and then construct a fusion representation vector based on these time-series feature variables and the time mapping function.

[0079] Step 102, performing spectrum analysis and enhancement on the multi-source timing information to obtain an oscillation signal containing a sub- / super-synchronous frequency band;

[0080] In this step, spectrum analysis and enhancement are mainly performed on the multi-source timing information to obtain an oscillation signal containing a sub- / super-synchronous frequency band.

[0081] In some embodiments, the implementation process of performing spectrum analysis and enhancement on the multi-source timing information to obtain an oscillation signal containing a sub- / super-synchronous frequency band can mainly include the following sub-steps S11 to S14:

[0082] Step S11: filtering the recording wave signal through a pre-constructed band-pass filter to obtain a sub- / super-synchronous frequency band;

[0083] The frequency band can be selected As a sub- / super-synchronous frequency band to be analyzed, a band-pass filter is constructed to filter out unnecessary frequency bands and focus on the sub- / super-synchronous frequency band. Among them, is the lower limit of the frequency of the frequency band, is the upper limit of the frequency of the frequency band. The band-pass filter satisfies the following amplitude-frequency response characteristics:

[0084]

[0085] Step S12: performing center frequency shift processing on the recording wave signal, and mapping the center frequency of the sub- / super-synchronous frequency band to a zero frequency point to obtain a frequency-shifted recording wave signal;

[0086] The recording wave signal sampling sequence is multiplied by a unit rotation factor to perform center frequency shift processing on the recording wave signal, map the center frequency of the sub- / super-synchronous frequency band to be analyzed to a zero frequency point, and obtain a frequency-shifted recording wave signal as follows:

[0087]

[0088] In the formula, is a unit imaginary number; and the center frequency is .

[0089] Step S13: resampling the frequency-shifted recording wave signal and performing spectrum analysis on the resampled signal through short-time Fourier transform;

[0090] The frequency-shifted recording wave signal is resampled, and the resampled signal is subjected to short-time Fourier transform to realize spectrum analysis.

[0091] Further, the step of performing the spectrum analysis based on the short-time Fourier transform comprises:

[0092] First, the resampled signal is segmented into a plurality of short-time window segments;

[0093] Next, Fourier transform is performed on each time window segment data to obtain the time-frequency spectrum of each time window segment .

[0094]

[0095] wherein, is a window function; is a frequency; is a time; and a Hanning window is selected as the window function of the Fourier transform. The expression of the window function is:

[0096]

[0097] wherein, .

[0098] Finally, the time-frequency spectrum of each time window segment is synthesized to obtain a time-frequency diagram, and the oscillation component in a specific frequency band is extracted from the time-frequency diagram to obtain the frequency result . The frequency result obtained according to the spectrum analysis based on the short-time Fourier transform can be used as the initial frequency for the subsequent identification operation.

[0099] Step S14: performing spectrum enhancement on the frequency-shifted recording wave signal according to the spectrum analysis result to obtain an oscillation signal in the sub- / super-synchronous frequency band.

[0100] The spectrum enhancement is performed on the frequency-shifted recording wave signal according to the spectrum analysis result of the short-time Fourier transform to improve the oscillation component in the sub- / super-synchronous frequency band, and an oscillation signal in the sub- / super-synchronous frequency band is obtained. The spectrum enhancement process mainly includes weighting adjustment on the low-energy part of the spectrum to increase the distinguishability of the oscillation signal.

[0101] Step 103: identifying the oscillation modal parameter of the oscillation signal in the sub- / super-synchronous frequency band based on the spectrum analysis result obtained in the spectrum analysis and enhancement process, and evaluating the oscillation risk level based on the oscillation modal parameter;

[0102] This step mainly realizes identification of the oscillation modal parameter of the oscillation signal in the sub- / super-synchronous frequency band based on the spectrum analysis result obtained in the spectrum analysis and enhancement process, and evaluation of the oscillation risk level based on the oscillation modal parameter.

[0103] In some embodiments, based on the spectrum analysis result, the oscillation mode parameter of the oscillation signal in the sub / super synchronous frequency band is identified, and the implementation process of evaluating the oscillation risk level based on the oscillation mode parameter can mainly include the following sub-steps S21-S25:

[0104] Step S21: A multi-modal signal model is constructed by considering the frequency, amplitude, damping coefficient and initial phase angle of the multi-modal oscillation at the same time.

[0105] The constructed multi-modal signal model is as follows:

[0106]

[0107] Wherein, m is the number of modes; is the amplitude of the mth mode; is the damping coefficient of the mth mode; is the frequency of the mth mode; is the initial phase angle of the mth mode (reflecting the phase trajectory). Step S22: A loss function is constructed by minimizing the square error between the oscillation signal and the model signal output by the multi-modal signal model; The constructed loss function is to minimize the square error between the observed signal (i.e. the oscillation signal) and the model signal (i.e. the model signal output by the model

[0108]

[0109] Wherein, y is the observed signal; is the model signal, represents the parameter combination to be solved for optimization.

[0110] Step S23: The constraint condition of the loss function is constructed based on the amplitude constraint, frequency constraint and damping ratio constraint;

[0111] The constructed constraint condition of the loss function is as follows:

[0112] Step S24: The loss function is optimized based on the constraint condition to obtain the oscillation mode parameter of the oscillation signal in the sub / super synchronous frequency band;

[0113] The optimization of the loss function based on the constraint condition is as follows:

[0114]

[0115] ​​​​​​Step S24: Obtain the spectrum analysis result in the spectrum analysis and enhancement process, extract the frequency of the spectrum analysis result as the initial frequency, and obtain the frequency, amplitude, initial phase angle and damping ratio of the multi-modal oscillation by optimizing the loss function under the constraint condition, as the oscillation modal parameters of the oscillation signal in the sub / super synchronous frequency band;

[0116] Under the constraint condition, the amplitude , frequency , damping ratio (reflecting the damping characteristic) and initial phase angle of the multi-modal oscillation can be obtained by optimizing the loss function.

[0117] Specifically, the optimization solving process of the loss function is as follows:

[0118] First, the frequency result of the spectrum analysis is taken as the initial value of optimization.

[0119] Then, the initial value of the frequency is substituted into the loss function, and the initial value of the decay coefficient is obtained by using one-dimensional optimization method.

[0120] Then, the initial value of the amplitude is obtained according to the linear least square solution.

[0121] Finally, the amplitude , frequency , damping ratio and initial phase angle of each modal oscillation are solved by loop minimization. Among them .

[0122] Step S25: Comprehensive analysis based on the frequency, amplitude, initial phase angle and damping ratio of the multi-modal oscillation, and evaluation of the oscillation risk level of the oscillation signal.

[0123] Comprehensive analysis based on the frequency, amplitude, initial phase angle and damping ratio of the multi-modal oscillation can evaluate the risk level of the oscillation. According to the oscillation fluctuation, the oscillation is divided into high risk, medium risk and low risk levels. Among them, the high risk oscillation is the oscillation with large oscillation amplitude and frequency close to the power grid frequency or the critical frequency of the device.

[0124] Step 104, according to the multi-source time sequence information and the oscillation modal parameter, the fusion representation vector is fused and verified, and the sub / super synchronous oscillation area and time sequence evolution path are identified;

[0125] According to the foregoing description, the multi-source time sequence information includes the synchronous phasor signal. This step mainly realizes fusion and verification of the fusion representation vector by the modal identification result (i.e., the oscillation modal parameter) obtained in the foregoing and the dynamic characteristics such as frequency drift, phase angle disturbance and power fluctuation in the synchronous phasor signal, to identify the possible sub- / ultra-synchronous oscillation region and the time sequence evolution path.

[0126] In some embodiments, the implementation process of fusion and verification of the fusion representation vector according to the multi-source time sequence information and the oscillation modal parameter to identify the sub- / ultra-synchronous oscillation region and the time sequence evolution path can mainly include the following sub-steps S31 to S36:

[0127] Step S31: Based on a preset sliding window length, time correlation of the oscillation modal parameter and the synchronous phasor signal under different time windows is calculated in combination with window sliding.

[0128] Specifically, the sliding window length and the overlap rate can be preset. In each window, the time correlation of each modal frequency trajectory of the oscillation modal parameter and the frequency change rate reflecting the frequency drift in the synchronous phasor signal, the phase angle velocity reflecting the phase angle disturbance, and the power fluctuation reflecting the power fluctuation condition is calculated.

[0129]

[0130] In the formula, corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; corr (f, f') represents the correlation of f and f'; , are weight factors for adjusting the influence of the corresponding item on the overall result.

[0131] Step S32: The target time window when the time correlation exceeds the preset correlation threshold is identified as the record starting point of the sub- / ultra-synchronous oscillation region, and the identification of the time sequence evolution path is triggered.

[0132] The criterion for triggering the identification of the oscillation region and the evolution path is that exceeds the set threshold. If exceeds the set threshold, the identification of the oscillation region is started, and the evolution path identification is triggered.

[0133] Step S33: Real-time transformation rate analysis is performed on the first derivative of the time correlation, and the evolution inflection point of the time sequence evolution path is judged based on the transformation rate analysis result;

[0134] Real-time transformation rate analysis is performed on the first derivative of the time correlation, If , it is determined that the oscillation path has an evolution inflection point. Wherein, is a pre-set threshold value.

[0135] In actual application, the threshold value may also be dynamically adjusted according to the oscillation intensity, frequency fluctuation, etc. of the system. When the system fluctuates violently (such as the frequency and power change rate are large), the threshold value can be appropriately increased to avoid misjudgment of the inflection point due to the fluctuation of the signal itself. When the system is stable, the threshold value can be appropriately reduced to enhance the sensitivity to weak changes and improve the identification sensitivity of the evolution inflection point.

[0136] Step S34: In the time sequence evolution path identification process, the fusion representation vector is fused and verified based on principal component analysis based on the oscillation modal parameter and the synchronous phasor signal, to obtain a principal component contribution degree matrix and a principal component time sequence change matrix;

[0137] Further, the fusion representation vector is fused and verified based on principal component analysis based on the oscillation modal parameter and the synchronous phasor signal, to obtain a principal component contribution degree matrix and a principal component time sequence change matrix, which can be realized by performing the following sub-steps S341 to S344:

[0138] Step S341: Construct a data matrix according to the time sequence of the oscillation modal parameter and the synchronous phasor signal, and perform standardization processing on the data matrix to obtain a feature matrix;

[0139] The modal frequency trajectory of the oscillation modal parameter ( , , ) and the multi-dimensional features of the synchronous phasor signal ( , , ) are constructed into a data matrix according to the time sequence. Wherein, each column represents a feature, and each row represents a time point. And the standardization processing can obtain a feature matrix .

[0140] Step S342: According to the feature matrix, a covariance matrix is constructed, and the covariance matrix is subjected to eigenvalue decomposition to obtain a plurality of eigenvalues and a feature vector corresponding to each eigenvalue;

[0141] According to the feature matrix , a covariance matrix The covariance matrix represents the variance and covariance between each pair of features, and its calculation formula is as follows:

[0142]

[0143] Where n represents the number of data points.

[0144] Next, the covariance matrix Eigenvalue decomposition yields multiple eigenvalues. and their respective feature vectors Eigenvalues This represents the contribution of each principal component to the data variability. The larger the eigenvalue, the greater the variance contained in that principal component, and thus the greater its contribution. Eigenvector The direction of the principal component is indicated, and the projection of the original features in the fused representation vector into the new space is defined.

[0145] Step S343: Sort the feature matrix by principal component contribution according to the order of eigenvalues ​​from largest to smallest, construct the principal component contribution matrix, and construct the eigenvector space matrix corresponding to the principal component contribution matrix based on each eigenvector;

[0146] Based on eigenvalues The principal components are ranked according to their contribution values ​​to obtain the principal component contribution matrix. The contribution of each principal component is determined by its corresponding eigenvalue. Determine the eigenvalues The larger the value, the greater the contribution. Then, based on each feature vector... Construct the eigenvector space matrix corresponding to the principal component contribution matrix. .

[0147] Furthermore, to reduce computational load and accelerate computation speed, principal components with higher contributions can be selected based on actual needs. Then, based on the selected principal components, the eigenvectors (and their corresponding principal components with high contributions) are combined to form an eigenvector space matrix. .

[0148] Step S344: Project the fused representation vector onto the feature vector space matrix to obtain the principal component temporal variation matrix.

[0149] Projecting the fused representation vector onto the eigenvector space matrix (corresponding to the selected principal component space) yields the temporal variation matrix of the principal components:

[0150]

[0151] in, Indicates the first The principal components at time 1 The projection value of the principal component reflects the performance of the oscillation signal in the principal component direction at this moment.

[0152] Step S35: Perform contribution analysis on the principal component contribution matrix to determine the final sub- / super-synchronous oscillation region.

[0153] By analyzing the contribution of the principal component, the final sub- / super-synchronous oscillation region is identified. A larger contribution of the principal component indicates a larger oscillation amplitude and a region where the system has a larger disturbance.

[0154] Step S36: According to the principal component time sequence change matrix, track the change trend of the oscillation mode over time, and determine the final time sequence evolution path according to the corresponding positions of the principal components at different times.

[0155] According to the time sequence change of the principal component , track the change trend of the oscillation mode over time, and identify the final time sequence evolution path according to the corresponding positions of the principal components at different times.

[0156] Step 105, reverse check the oscillation mode parameters, when the reverse check is passed, output the oscillation mode parameters, the oscillation risk level, the sub- / super-synchronous oscillation region and the time sequence evolution path as the multi-dimensional recognition result of the oscillation signal.

[0157] In combination with the foregoing discussion, the multi-source time sequence information also includes a protection event action signal. In this step, based on the trigger time of the protection event action signal as an anchor point, the oscillation mode parameter recognition result is reverse verified within a certain window range before the event trigger to determine whether the multi-modal trajectory of the oscillation signal has a causal relationship with the protection action trigger event. When it is determined that there is an association, the multi-dimensional recognition result including the oscillation mode parameters, the time sequence evolution path (reflecting the time-varying trajectory), the oscillation risk level and the possible triggered sub- / super-synchronous oscillation region is output.

[0158] In some embodiments, the implementation process of reverse checking the oscillation mode parameters, when the reverse check is passed, outputting the oscillation mode parameters, the oscillation risk level, the sub- / super-synchronous oscillation region and the time sequence evolution path as the multi-dimensional recognition result of the oscillation signal, can mainly include the following sub-steps S41 to S43:

[0159] Step S41: Take the trigger time of the protection event action signal as an anchor point, and construct a time backtracking window based on the trigger time;

[0160] Let the trigger time of the protection event action signal be , and the backtracking window be .

[0161] Step S42: Calculate the power spectrum energy density of the oscillation mode parameters in the time backtracking window;

[0162] The power spectrum energy density of the identified modalities in the time period is calculated by the following formula :

[0163]

[0164] Step S43: When the power spectrum energy density is greater than the preset energy density threshold and the multi-modal oscillation frequency is in the sub- / super-synchronous frequency band, it is determined that the multi-modal trajectory of the oscillation signal is associated with the protection action, and the oscillation modal parameters, the oscillation risk level, the sub- / super-synchronous oscillation region and the time sequence evolution path are output as the multi-dimensional identification result of the oscillation signal.

[0165] If and the corresponding , the multi-modal trajectory of the oscillation signal is considered as the modality associated with the protection action, and the final result output step is entered.

[0166] If the multi-modal trajectory of the oscillation signal has an energy overlap region near the time , the composite energy function is defined as:

[0167]

[0168] Wherein, is the modal weight function, which is related to the consistency degree of the synchronous phasor signal feature.

[0169] Finally, when the reverse verification passes, the oscillation modal parameters, the oscillation risk level, the sub- / super-synchronous oscillation region and the time sequence evolution path are output as the multi-dimensional identification result of the oscillation signal.

[0170] In the embodiment of the present application, a power system sub / super-synchronous oscillation signal accurate identification method based on multi-source data fusion is proposed. First, by organically integrating the monitored multi-type data sources, the identification of the oscillation signal is more comprehensive and robust, and the time alignment and fusion modeling between the data can also be achieved, providing a unified foundation for subsequent analysis. Then, by using spectrum analysis and enhancement methods, not only the oscillation signal containing the sub / super-synchronous frequency band can be obtained, but also the identification sensitivity to weak signals can be significantly improved, which is conducive to discovering potential system instability factors in the early stage. Then, based on the spectrum analysis results obtained in the spectrum analysis and enhancement process, the oscillation modal parameters of the oscillation signal in the sub / super-synchronous frequency band are identified by optimization, and the oscillation risk level is further evaluated based on the oscillation modal parameters, so as to evaluate the current risk situation of the system, so that the high-risk situation can be responded in time. Then, the multi-modal oscillation parameters and the multi-dimensional dynamic characteristics of the synchronous phasor signal are combined to realize fusion verification, identify the sub / super-synchronous oscillation region and time sequence evolution path, which can improve the accuracy and reliability of the identification result and reduce the risk of false alarm and missed alarm. Finally, the reverse verification based on the causal correlation analysis is carried out combined with the protection action, and when the reverse verification passes, the oscillation modal parameters, the oscillation risk level, the sub / super-synchronous oscillation region and the time sequence evolution path are output as the multi-dimensional identification result of the oscillation signal, which has practical regulation significance.

[0171] For better illustration, refer to Figure 2 , a whole flowchart of a power system sub / super-synchronous oscillation signal identification method provided by an embodiment of the present application is shown. It should be pointed out that this embodiment only briefly describes the general process of power system sub / super-synchronous oscillation signal identification, and the specific implementation process of each step can be understood by referring to the related contents in the foregoing embodiments, which will not be described here again. It can be understood that the present application does not limit this.

[0172] Step 201: Collecting multi-source time sequence information of the power system, and performing time alignment and time reconstruction on the multi-source time sequence information, and then constructing a fusion representation vector through feature extraction and time mapping;

[0173] Step 202: Based on the center frequency shift processing and the short-time Fourier transform, the multi-source time sequence information is subjected to spectrum analysis and enhancement, and the oscillation signal containing the sub / super-synchronous frequency band is obtained;

[0174] Step 203: Based on the spectrum analysis results obtained in the spectrum analysis and enhancement process, the oscillation modal parameters of the oscillation signal in the sub / super-synchronous frequency band are identified by optimization, and the oscillation risk level is evaluated based on the oscillation modal parameters;

[0175] Step 204: performing principal component analysis-based fusion verification analysis on the fusion representation vector according to the multi-source time sequence information and the oscillation mode parameter, identifying a sub / super-synchronous oscillation region and a time sequence evolution path;

[0176] Step 205: performing reverse verification on the oscillation mode parameter based on protection action correlation analysis;

[0177] Step 206: when the reverse verification passes, outputting the oscillation mode parameter, the oscillation risk level, the sub / super-synchronous oscillation region and the time sequence evolution path as a multi-dimensional identification result of the oscillation signal.

[0178] Referring to Figure 3 , a structural block diagram of an oscillation signal identification device for a power system is shown, which can specifically include:

[0179] A fusion representation vector construction unit 301 is configured to collect multi-source time sequence information of the power system, and construct a fusion representation vector according to the multi-source time sequence information;

[0180] A spectrum analysis and enhancement unit 302 is configured to perform spectrum analysis and enhancement on the multi-source time sequence information, and obtain an oscillation signal containing a sub / super-synchronous frequency band;

[0181] An oscillation mode parameter identification unit 303 is configured to identify an oscillation mode parameter of the oscillation signal in the sub / super-synchronous frequency band based on a spectrum analysis result obtained in the spectrum analysis and enhancement process, and evaluate an oscillation risk level based on the oscillation mode parameter;

[0182] A fusion verification unit 304 is configured to perform fusion verification on the fusion representation vector according to the multi-source time sequence information and the oscillation mode parameter, and identify a sub / super-synchronous oscillation region and a time sequence evolution path;

[0183] A reverse verification unit 305 is configured to perform reverse verification on the oscillation mode parameter, and when the reverse verification passes, output the oscillation mode parameter, the oscillation risk level, the sub / super-synchronous oscillation region and the time sequence evolution path as a multi-dimensional identification result of the oscillation signal.

[0184] In an optional embodiment, the multi-source time sequence information includes a recorded wave signal, a synchrophasor signal, a protection event action signal and a device state quantity; and the fusion representation vector construction unit 301 includes:

[0185] A time alignment and reconstruction unit is configured to perform time alignment and time reconstruction on the recorded wave signal, the synchrophasor signal, the protection event action signal and the device state quantity with the timestamp of the synchrophasor signal as a reference axis, and construct a plurality of multi-source data corresponding to different time periods;

[0186] a time mapping unit configured to extract a time sequence characteristic variable of the multi-source data in each time period, and perform time mapping based on the time sequence characteristic variable to construct a fusion representation vector.

[0187] In an optional embodiment, the multi-source time sequence information comprises a recorded wave signal; and the spectrum analysis and enhancement unit 302 comprises:

[0188] a filtering processing unit configured to perform filtering processing on the recorded wave signal by using a pre-constructed band-pass filter to obtain a sub- / super-synchronous frequency band;

[0189] a center frequency shift processing unit configured to perform center frequency shift processing on the recorded wave signal, and map a center frequency of the sub- / super-synchronous frequency band to a zero frequency point to obtain a frequency-shifted recorded wave signal;

[0190] a spectrum analysis unit configured to perform resampling on the frequency-shifted recorded wave signal, and perform spectrum analysis on the resampled signal by using a short-time Fourier transform;

[0191] a spectrum enhancement unit configured to perform spectrum enhancement on the frequency-shifted recorded wave signal according to the spectrum analysis result to obtain an oscillation signal in the sub- / super-synchronous frequency band.

[0192] In an optional embodiment, the oscillation modal parameter identification unit 303 comprises:

[0193] a multi-modal signal model construction unit configured to construct a multi-modal signal model by simultaneously considering frequency, amplitude, damping coefficient and initial phase angle of multi-modal oscillation;

[0194] a loss function construction unit configured to construct a loss function by taking minimizing a square error between the oscillation signal and a model signal output by the multi-modal signal model as an objective;

[0195] a constraint condition construction unit configured to construct a constraint condition of the loss function based on amplitude constraint, frequency constraint and damping ratio constraint;

[0196] an optimization solving unit configured to obtain a spectrum analysis result in a spectrum analysis and enhancement process, extract a frequency of the spectrum analysis result as an initial frequency, and obtain frequency, amplitude, initial phase angle and damping ratio of multi-modal oscillation as oscillation modal parameters of the oscillation signal in the sub- / super-synchronous frequency band by optimizing and solving the loss function under the constraint condition;

[0197] an oscillation risk level evaluation unit configured to comprehensively analyze the frequency, amplitude, initial phase angle and damping ratio of the multi-modal oscillation to evaluate an oscillation risk level of the oscillation signal.

[0198] In an optional embodiment, the multi-source timing information comprises a synchrophasor signal; and the fusion verification unit 304 comprises:

[0199] a time correlation calculation unit configured to calculate time correlation between the oscillation modal parameter and the synchrophasor signal in different time windows based on a preset sliding window length and in combination with window sliding;

[0200] a timing path identification triggering unit configured to identify a target time window when the time correlation exceeds a preset correlation threshold as a recording starting point of a sub- / super-synchronous oscillation region, and trigger identification of a timing evolution path;

[0201] a transformation rate analysis unit configured to perform transformation rate analysis on a first derivative of the time correlation in real time, and determine an evolution inflection point of the timing evolution path based on a result of the transformation rate analysis;

[0202] a fusion verification analysis unit configured to perform principal component analysis-based fusion verification analysis on the fusion representation vector based on the oscillation modal parameter and the synchrophasor signal during identification of the timing evolution path, to obtain a principal component contribution degree matrix and a principal component timing change matrix;

[0203] a contribution degree analysis unit configured to perform contribution degree analysis on the principal component contribution degree matrix, to determine a final sub- / super-synchronous oscillation region;

[0204] a timing evolution path determination unit configured to track a change trend of an oscillation mode over time according to the principal component timing change matrix, and determine a final timing evolution path according to positions of different principal components.

[0205] In an optional embodiment, the fusion verification analysis unit comprises:

[0206] a feature matrix construction unit configured to construct a data matrix from the oscillation modal parameter and the synchrophasor signal in time sequence, and perform standardization processing on the data matrix to obtain a feature matrix;

[0207] an eigenvalue decomposition unit configured to construct a covariance matrix from the feature matrix, and perform eigenvalue decomposition on the covariance matrix to obtain a plurality of eigenvalues and a feature vector corresponding to each of the eigenvalues;

[0208] a principal component contribution degree sorting unit configured to sort the feature matrix according to an order of eigenvalues from large to small, to construct a principal component contribution degree matrix, and to construct a feature vector space matrix corresponding to the principal component contribution degree matrix based on each of the feature vectors;

[0209] a projection unit configured to project the fusion representation vector to the feature vector space matrix to obtain a principal component timing change matrix.

[0210] In an optional embodiment, the multi-source timing information comprises a protection event action signal; the reverse check unit 305 comprises:

[0211] a time backtracking window construction unit, configured to take a trigger time of the protection event action signal as an anchor point, and construct a time backtracking window based on the trigger time;

[0212] a power spectrum energy density calculation unit, configured to calculate a power spectrum energy density of the oscillation modal parameter in the time backtracking window;

[0213] a recognition result output unit, configured to determine that a multi-modal trajectory of the oscillation signal is associated with a protection action when the power spectrum energy density is greater than a preset energy density threshold and a multi-modal oscillation frequency is located in the sub- / ultra-synchronous frequency band, and output the oscillation modal parameter, the oscillation risk level, the sub- / ultra-synchronous oscillation region and the timing evolution path as a multi-dimensional recognition result of the oscillation signal.

[0214] For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the related parts can be referred to the foregoing description of the method embodiment.

[0215] The embodiment of the present application further provides an electronic device, which comprises a processor and a memory:

[0216] The memory is used for storing program code and transmitting the program code to the processor;

[0217] The processor is used for executing the power system sub- / ultra-synchronous oscillation signal recognition method according to the instructions in the program code.

[0218] The embodiment of the present application further provides a computer readable storage medium, which is used for storing program code, and the program code is used for executing the power system sub- / ultra-synchronous oscillation signal recognition method.

[0219] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiment, and will not be repeated here.

[0220] In several embodiments provided by the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the described device embodiments are merely schematic. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0221] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0222] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can be physically present separately, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0223] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part of the prior art that contributes to the technical solutions or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0224] The above embodiments are merely used to illustrate the technical solutions of the present application, rather than limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent replacements to some technical features. Such modifications or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for identifying subsynchronous / supersynchronous oscillation signals in a power system, characterized in that, include: Collect multi-source time-series information of the power system, including waveform recording signals, synchronization phasor signals, protection event action signals, and equipment status quantities; The fusion representation vector is constructed based on the multi-source time-series information, including: using the timestamp of the synchronization phasor signal as a reference axis, performing time alignment and time reconstruction on the waveform recording signal, the synchronization phasor signal, the protection event action signal, and the device status quantity to construct multiple multi-source data corresponding to different time periods; for the multi-source data in each time period, extracting the time-series feature variables of the multi-source data, and performing time mapping based on the time-series feature variables to construct the fusion representation vector; The multi-source timing information is subjected to spectral analysis and enhancement to obtain an oscillation signal containing sub / supersynchronous frequency bands; Based on the spectrum analysis results obtained during the spectrum analysis and enhancement process, the oscillation mode parameters of the oscillation signal in the sub / supersynchronous frequency band are identified, and the oscillation risk level is assessed based on the oscillation mode parameters; The process involves: performing fusion verification on the fused representation vector based on the multi-source time series information and the oscillation mode parameters; identifying sub / supersynchronous oscillation regions and time series evolution paths; including: calculating the time correlation between the oscillation mode parameters and the synchronization phasor signal under different time windows based on a preset sliding window length; identifying the target time window where the time correlation exceeds a preset correlation threshold as the recording start point of the sub / supersynchronous oscillation region and triggering the identification of the time series evolution path; performing transformation rate analysis on the first derivative of the time correlation in real time; determining the evolution inflection point of the time series evolution path based on the transformation rate analysis results; performing fusion verification analysis on the fused representation vector based on principal component analysis based on the oscillation mode parameters and the synchronization phasor signal during the time series evolution path identification process; obtaining the principal component contribution matrix and the principal component time series change matrix; performing contribution analysis on the principal component contribution matrix to determine the final sub / supersynchronous oscillation region; tracking the change trend of the oscillation mode over time based on the principal component time series change matrix; and determining the final time series evolution path based on the corresponding positions of the principal components at different times. The oscillation mode parameters are reverse-verified. When the reverse verification passes, the oscillation mode parameters, the oscillation risk level, the sub / supersynchronous oscillation region, and the timing evolution path are output as the multidimensional identification result of the oscillation signal. This includes: using the trigger time of the protection event action signal as the anchor point, constructing a time backtracking window based on the trigger time, calculating the power spectral energy density of the oscillation mode parameters within the time backtracking window, and when the power spectral energy density is greater than a preset energy density threshold and the multimodal oscillation frequency is within the sub / supersynchronous frequency band, determining that the multimodal trajectory of the oscillation signal is associated with the protection action, and outputting the oscillation mode parameters, the oscillation risk level, the sub / supersynchronous oscillation region, and the timing evolution path as the multidimensional identification result of the oscillation signal.

2. The method for identifying sub / supersynchronous oscillation signals in power systems according to claim 1, characterized in that, The multi-source timing information includes recorded waveforms; the step of performing spectral analysis and enhancement on the multi-source timing information to obtain an oscillation signal containing sub / supersynchronous frequency bands includes: The recorded signal is filtered by a pre-constructed bandpass filter to obtain the sub / supersynchronous frequency band; The recorded waveform signal is subjected to center frequency shifting processing, and the center frequency of the sub / supersynchronous frequency band is mapped to the zero frequency point to obtain the frequency shifted recorded waveform signal; The frequency-shifted waveform signal is resampled, and the resampled signal is subjected to spectrum analysis by short-time Fourier transform. Based on the spectral analysis results, the frequency-shifted waveform signal is enhanced to obtain the oscillation signal of the sub / supersynchronous frequency band.

3. The method for identifying sub / supersynchronous oscillation signals in power systems according to claim 1, characterized in that, Based on the spectrum analysis results obtained during the spectrum analysis and enhancement process, the oscillation mode parameters of the oscillation signal within the sub / supersynchronous frequency band are identified, and the oscillation risk level is assessed based on the oscillation mode parameters, including: Simultaneously considering the frequency, amplitude, attenuation coefficient, and initial phase angle of multimodal oscillations, a multimodal signal model is constructed; A loss function is constructed with the objective of minimizing the squared error between the oscillation signal and the output model signal of the multimodal signal model; Based on amplitude constraints, frequency constraints, and damping ratio constraints, the constraints of the loss function are constructed. The spectrum analysis results are obtained during the spectrum analysis and enhancement process. The frequency of the spectrum analysis results is extracted as the initial frequency. The loss function is optimized and solved under the constraints to obtain the frequency, amplitude, initial phase angle and damping ratio of the multimode oscillation, which are used as the oscillation mode parameters of the oscillation signal in the sub / supersynchronous frequency band. Based on a comprehensive analysis of the frequency, amplitude, initial phase angle, and damping ratio of the multimodal oscillation, the oscillation risk level of the oscillation signal is assessed.

4. The method for identifying sub / supersynchronous oscillation signals in power systems according to claim 1, characterized in that, The step of performing a fusion verification analysis based on principal component analysis on the fused representation vector using the oscillation mode parameters and the synchronization phasor signal to obtain the principal component contribution matrix and the principal component temporal variation matrix includes: The oscillation mode parameters and the synchronization phasor signal are used to construct a data matrix according to the time series, and the data matrix is ​​standardized to obtain the feature matrix; Construct a covariance matrix based on the feature matrix, and perform eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues ​​and eigenvectors corresponding to each eigenvalue; The feature matrix is ​​sorted by principal component contribution according to the eigenvalues ​​from largest to smallest to construct a principal component contribution matrix, and an eigenvector space matrix corresponding to the principal component contribution matrix is ​​constructed based on each eigenvector. The fused representation vector is projected onto the feature vector space matrix to obtain the principal component temporal variation matrix.

5. A device for identifying sub / supersynchronous oscillation signals in a power system, characterized in that, include: A fusion representation vector construction unit is used to collect multi-source time-series information of the power system, including waveform recording signals, synchronization phasor signals, protection event action signals, and equipment status quantities. The fusion representation vector is constructed based on the multi-source time-series information, including: using the timestamp of the synchronization phasor signal as a reference axis, performing time alignment and time reconstruction on the waveform recording signal, the synchronization phasor signal, the protection event action signal, and the device status quantity to construct multiple multi-source data corresponding to different time periods; for the multi-source data in each time period, extracting the time-series feature variables of the multi-source data, and performing time mapping based on the time-series feature variables to construct the fusion representation vector; The spectrum analysis and enhancement unit is used to perform spectrum analysis and enhancement on the multi-source timing information to obtain an oscillation signal containing sub / supersynchronous frequency bands; The oscillation mode parameter identification unit is used to identify the oscillation mode parameters of the oscillation signal in the sub / supersynchronous frequency band based on the spectrum analysis results obtained during the spectrum analysis and enhancement process, and to assess the oscillation risk level based on the oscillation mode parameters; The fusion verification unit is used to perform fusion verification on the fusion representation vector based on the multi-source time series information and the oscillation mode parameters, and to identify the sub / supersynchronous oscillation region and time series evolution path. This includes: calculating the time correlation between the oscillation mode parameters and the synchronization phasor signal under different time windows based on a preset sliding window length, identifying the target time window where the time correlation exceeds a preset correlation threshold as the recording start point of the sub / supersynchronous oscillation region, and triggering the identification of the time series evolution path; performing real-time transformation rate analysis on the first derivative of the time correlation, and determining the evolution inflection point of the time series evolution path based on the transformation rate analysis results; during the time series evolution path identification process, performing fusion verification analysis based on principal component analysis on the fusion representation vector based on the oscillation mode parameters and the synchronization phasor signal to obtain the principal component contribution matrix and the principal component time series change matrix; performing contribution analysis on the principal component contribution matrix to determine the final sub / supersynchronous oscillation region; tracking the change trend of the oscillation mode over time based on the principal component time series change matrix; and determining the final time series evolution path based on the corresponding positions of the principal components at different times. A reverse verification unit is used to perform reverse verification on the oscillation mode parameters. When the reverse verification passes, it outputs the oscillation mode parameters, the oscillation risk level, the sub / supersynchronous oscillation region, and the timing evolution path as the multidimensional identification result of the oscillation signal. This includes: using the trigger time of the protection event action signal as an anchor point, constructing a time backtracking window based on the trigger time, calculating the power spectral energy density of the oscillation mode parameters within the time backtracking window, and when the power spectral energy density is greater than a preset energy density threshold and the multimodal oscillation frequency is within the sub / supersynchronous frequency band, determining that the multimodal trajectory of the oscillation signal is associated with the protection action, and outputting the oscillation mode parameters, the oscillation risk level, the sub / supersynchronous oscillation region, and the timing evolution path as the multidimensional identification result of the oscillation signal.

6. An electronic device, characterized in that, The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the power system subsynchronous / supersynchronous oscillation signal identification method according to any one of claims 1-4, based on the instructions in the program code.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store program code for executing the power system sub / supersynchronous oscillation signal identification method according to any one of claims 1-4.

Images