Low-frequency oscillation identification method and system based on PMU data

Abstract

The invention discloses a low-frequency oscillation identification method and system based on PMU data, and belongs to the technical field of power system operation monitoring, and the method comprises the steps: collecting PMU data from a synchronous phasor measurement device through a communication interface; establishing a global data buffer area, and setting a multi-resolution analysis window covering different oscillation frequency bands according to a PMU sampling rate; performing peak value and valley value detection on the data in each analysis window, setting a neighborhood minimum distance parameter to identify extreme points according to the type of each window, and generating a corresponding extreme point sequence; calculating oscillation frequency and oscillation amplitude according to the extreme point sequence; and judging the oscillation state according to the oscillation frequency, the oscillation amplitude and the noise level obtained by each analysis window, and outputting a detection result. According to the method, the false detection rate is reduced from 30% to 5% or below through dynamic noise estimation, the omission ratio is reduced from 25% to 3% or below, and the method can adapt to complex noise scenes such as new energy output fluctuation and load impact.

Description

[0001] This invention relates to the field of power system operation monitoring technology, specifically to a method and system for identifying low-frequency oscillations based on PMU data. Background Technology

[0002] With the advancement of new power system construction, distributed power sources, energy storage devices, and power electronic equipment are being connected on a large scale, significantly changing the operating characteristics of power plants. Low-frequency oscillations in the power system are a typical risk in the operation of grids with high proportions of new energy integration and inter-regional interconnection. Their frequency typically ranges from 0.1 to 2.5 Hz. If not detected and intervened in a timely manner, they can lead to voltage / frequency instability, generator disconnection, or even large-scale power outages. Phasor measurement units (PMUs), with their millisecond-level time synchronization accuracy and high sampling rate, have become the core hardware support for real-time monitoring of low-frequency oscillations. However, the extraction of their data value is limited by the shortcomings of existing detection methods, specifically including: 1) Existing methods mostly use a single fixed-length window to analyze PMU data. If the window is too long, although it can improve the frequency resolution of 0.1~2.5Hz oscillations, the response delay for 2~2.5Hz oscillations exceeds 2 seconds, missing the best intervention opportunity. If the window is too short, the oscillation response delay is reduced to less than 0.8 seconds, but the oscillation frequency error exceeds 0.2Hz, which is easy to misjudge as "no oscillation" or "frequency shift".

[0003] 2) Conventional detection methods use a fixed neighborhood distance (such as 5 sampling points) to identify signal extreme points. When facing the "rapid fluctuation" characteristics of high-frequency oscillations, extreme points are easily missed due to the neighborhood being too wide. When facing the "slow change" characteristics of low-frequency oscillations, noise extreme points are easily falsely detected due to the neighborhood being too narrow, resulting in a frequency calculation deviation of more than 10%.

[0004] 3) The power system has noise such as load fluctuations and disturbances from new energy output (the amplitude is usually 0.5% to 2% of the fundamental frequency). Existing methods use fixed thresholds to determine oscillations (such as amplitude exceeding the fundamental frequency by 1%). When noise is superimposed, the false detection rate exceeds 30%, and when the noise is small, the missed detection rate exceeds 25%, which cannot dynamically adapt to complex operating environments.

[0005] 4) The PMU output data contains strict time stamps and multi-dimensional information. Existing methods do not make full use of the time stamp synchronization characteristics and the multi-dimensional data fusion is insufficient, which makes it impossible to compare the oscillation detection results of different PMU measurement points horizontally and makes it difficult to locate the oscillation source.

[0006] To address the aforementioned issues, a low-frequency oscillation detection method that is deeply adapted to the characteristics of the PMU, can dynamically balance accuracy and real-time performance, and has strong noise immunity is needed to provide reliable technical support for the stable operation of the power grid. Summary of the Invention

[0007] In view of the above-mentioned problems, the present invention is proposed.

[0008] Therefore, the technical problem solved by this invention is that a synchronous phasor measurement unit (PMU) can provide measured quantities such as voltage, phase angle, and frequency with precise time scales, providing a data foundation for low-frequency oscillation detection. However, existing detection methods fail to effectively utilize the high-precision synchronization characteristics of the PMU and cannot resolve the contradiction between detection accuracy and real-time performance.

[0009] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a low-frequency oscillation identification method based on PMU data, comprising: acquiring PMU data from a synchronous phasor measurement device through a communication interface; performing packet loss detection and time stamp correction on the data; filtering out invalid frame data; and obtaining a valid PMU data sequence. A global data buffer is established to store the valid PMU data; a multi-resolution analysis window covering different oscillation frequency bands is set according to the PMU sampling rate, and the latest data of window 1, window 2 and window 3 are extracted from the buffer; Peak and valley detection is performed on the data within each analysis window. The minimum neighborhood distance parameter is set according to each window type to identify extreme points and generate the corresponding extreme point sequence. The oscillation frequency and oscillation amplitude are calculated based on the extreme point sequence, a differential signal is calculated on the effective PMU data to obtain the noise level, and the oscillation judgment threshold is adjusted according to the noise level. The oscillation state is determined based on the oscillation frequency, oscillation amplitude, and noise level obtained from each analysis window, and the detection results are output.

[0010] As a preferred embodiment of the low-frequency oscillation identification method based on PMU data described in this invention, the multi-resolution analysis window includes window 1, window 2 and window 3; Window 1 corresponds to the detection frequency of oscillation signals from 0.1 to 0.5 Hz, window 2 corresponds to the detection frequency of oscillation signals from 0.5 to 2.0 Hz, and window 3 corresponds to the detection frequency of oscillation signals from 2.0 to 2.5 Hz. The length of each window is determined based on the preset number of cycles and sampling rate, forming detection windows for oscillation signals in different frequency bands.

[0011] As a preferred embodiment of the low-frequency oscillation identification method based on PMU data described in this invention, wherein: when performing extreme point detection, a minimum neighborhood distance parameter is set for different analysis windows. and ; Window 1 uses neighborhood distance ; Window 2 uses neighborhood distance ; Window 3 uses neighborhood distance ; During the extreme value detection process, false detection points with adjacent spacing less than the neighborhood distance are eliminated based on the neighborhood distance.

[0012] As a preferred embodiment of the low-frequency oscillation identification method based on PMU data described in this invention, the oscillation frequency is calculated based on the time interval between adjacent extreme points; Window 1 calculates the frequency using the average time interval of the most recent four cycles; Window 2 uses the average time interval over the entire period to calculate the frequency; Window 3 calculates the frequency using the average time interval of the two most recent cycles.

[0013] As a preferred embodiment of the low-frequency oscillation identification method based on PMU data described in this invention, the oscillation amplitude is determined according to the maximum peak value at the extreme point. and minimum valley value Calculate the amplitude ; When the number of extreme points is less than three sets, the peak value of the window data is used for amplitude calculation.

[0014] As a preferred embodiment of the low-frequency oscillation identification method based on PMU data described in this invention, wherein: the differential signal between adjacent sampling points is calculated from the valid PMU data. ; Calculate the noise level based on the average value of the differential signal. The oscillation detection threshold is updated based on the noise level.

[0015] As a preferred embodiment of the low-frequency oscillation identification method based on PMU data described in this invention, the oscillation state determination is based on the joint conditions of the validity of the oscillation frequency, confidence level, signal-to-noise ratio, and amplitude ratio. When satisfied When the confidence level is greater than 0.3, the signal-to-noise ratio is greater than 2.0, and the amplitude percentage is not less than 0.5%, an oscillation state is determined to exist, and the oscillation frequency, oscillation amplitude, and confidence level are output.

[0016] This invention provides a system for identifying low-frequency oscillations based on PMU data.

[0017] To solve the above technical problems, the present invention provides the following technical solution: a system for identifying low-frequency oscillations based on PMU data, comprising: a data acquisition module, used to acquire PMU data from a synchronous phasor measurement device through a communication interface, perform packet loss detection and time stamp correction on the data, filter out invalid frame data, and obtain a valid PMU data sequence; The data buffer module is used to establish a global data buffer to store the effective PMU data, set a multi-resolution analysis window covering different oscillation frequency bands according to the PMU sampling rate, and extract the latest data of window 1, window 2 and window 3 from the buffer. The extreme value detection module is used to perform peak and valley value detection on the data in each analysis window, identify extreme points according to the minimum neighborhood distance parameter set according to each window type, and generate the corresponding extreme point sequence. The parameter calculation module is used to calculate the oscillation frequency and oscillation amplitude based on the extreme point sequence, calculate the differential signal of the effective PMU data to obtain the noise level, and adjust the oscillation judgment threshold according to the noise level. The status output module is used to determine the oscillation status based on the oscillation frequency, oscillation amplitude and noise level obtained from each analysis window, and output the detection results.

[0018] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the low-frequency oscillation identification method based on PMU data.

[0019] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the low-frequency oscillation identification method based on PMU data.

[0020] The beneficial effects of this invention: Through multi-window differentiated design, low-frequency oscillation ( (Frequency error <0.03Hz, response delay <3 seconds; high-frequency oscillation 2.0-2.5Hz) response delay <0.6 seconds, error <0.15Hz. Compared with the fixed window method, the accuracy is improved by 40% and the real-time performance is improved by 60%.

[0021] 3 types of window coverage It covers the entire frequency band, avoiding the inherent contradiction of existing methods that have fast high-frequency response but poor low-frequency accuracy or high low-frequency accuracy but slow high-frequency response, achieving 100% oscillation detection coverage.

[0022] Dynamic noise estimation reduces the false detection rate from 30% to below 5% and the false negative rate from 25% to below 3%, making it suitable for complex noise scenarios such as fluctuations in new energy output and load shocks.

[0023] It fully utilizes the time-stamping synchronization characteristics of PMU to support the comparison of oscillation results at multiple measurement points, providing a basis for oscillation source localization, and the point-by-point real-time processing mode is adapted to the high sampling rate of PMU.

[0024] It provides complete detection status information and confidence level assessment, making it easy to integrate into existing monitoring systems. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 The above is a flowchart of a low-frequency oscillation identification method based on PMU data, provided as an embodiment of the present invention. Detailed Implementation

[0027] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0028] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a low-frequency oscillation identification method based on PMU data, including: PMU data is acquired from the synchronous phasor measurement device through the communication interface, and packet loss detection and time stamp correction are performed on the data to filter out invalid frame data and obtain a valid PMU data sequence. A global data buffer is established to store the valid PMU data; a multi-resolution analysis window covering different oscillation frequency bands is set according to the PMU sampling rate, and the latest data of window 1, window 2 and window 3 are extracted from the buffer; Peak and valley detection is performed on the data within each analysis window. The minimum neighborhood distance parameter is set according to each window type to identify extreme points and generate the corresponding extreme point sequence. The oscillation frequency and oscillation amplitude are calculated based on the extreme point sequence, a differential signal is calculated on the effective PMU data to obtain the noise level, and the oscillation judgment threshold is adjusted according to the noise level. The oscillation state is determined based on the oscillation frequency, oscillation amplitude, and noise level obtained from each analysis window, and the detection results are output.

[0029] The multi-resolution analysis window includes window 1, window 2, and window 3; Window 1 corresponds to the detection frequency of oscillation signals from 0.1 to 0.5 Hz, window 2 corresponds to the detection frequency of oscillation signals from 0.5 to 2.0 Hz, and window 3 corresponds to the detection frequency of oscillation signals from 2.0 to 2.5 Hz. The length of each window is determined based on the preset number of cycles and sampling rate, forming detection windows for oscillation signals in different frequency bands.

[0030] When performing extreme point detection, the minimum neighborhood distance parameter is set separately for different analysis windows. and ; Window 1 uses neighborhood distance ; Window 2 uses neighborhood distance ; Window 3 uses neighborhood distance ; During the extreme value detection process, false detection points with adjacent spacing less than the neighborhood distance are eliminated based on the neighborhood distance.

[0031] The oscillation frequency is calculated based on the time interval between adjacent extreme points; Window 1 calculates the frequency using the average time interval of the most recent four cycles; Window 2 uses the average time interval over the entire period to calculate the frequency; Window 3 calculates the frequency using the average time interval of the two most recent cycles.

[0032] The oscillation amplitude is based on the maximum peak value at the extreme point. and minimum valley value Calculate the amplitude ; When the number of extreme points is less than three sets, the peak value of the window data is used for amplitude calculation.

[0033] Calculate the differential signal between adjacent sampling points for the valid PMU data. ; Calculate the noise level based on the average value of the differential signal. The oscillation detection threshold is updated based on the noise level.

[0034] The oscillation state determination is based on a combination of the validity of the oscillation frequency, confidence level, signal-to-noise ratio, and amplitude proportion. When satisfied When the confidence level is greater than 0.3, the signal-to-noise ratio is greater than 2.0, and the amplitude percentage is not less than 0.5%, an oscillation state is determined to exist, and the oscillation frequency, oscillation amplitude, and confidence level are output.

[0035] This embodiment of the method achieves high-precision real-time detection of low-frequency oscillations driven by PMU data through a fusion strategy of multi-window parallel analysis and adaptive extreme value detection combined with dynamic noise estimation. It resolves the contradiction between detection accuracy and frequency resolution, meets the real-time monitoring requirements of power systems, and is applicable to real-time monitoring of low-frequency oscillations in the 0.1~2.5Hz frequency band. It can support application scenarios such as power grid stability control, fault early warning, and operation status assessment.

[0036] Example 2, an embodiment of the present invention, provides a low-frequency oscillation identification method based on PMU data, based on the previous embodiment, including: Step 1: PMU Data Acquisition and Preprocessing. Connect the PMU device via Ethernet or fiber optic interface to acquire time-stamped data output by the PMU. Prioritize active power or voltage phase angle as the detection signal, filtering out invalid frames such as packet loss and abnormal time-stamped frames from the PMU data. Trigger an alarm when the proportion of invalid frames exceeds 10%. Extract the pulse-per-second (PPS) signal from the PMU data, ensuring that the timestamps of all sampling points are synchronized with the global clock, with a time deviation of ≤1ms between different PMU measurement points.

[0037] Step 2: Create a global data buffer (based on a double-ended queue, maximum length = , ), store the most recent Seconds of PMU data; when the buffer data size is ≥ Extract the latest data for each of the three types of windows: (1) Window 1 data: last buffer Points: (2) Window 2 data: last buffer One point; (3) Window 3 data: last buffer One point.

[0038] Step 3: Multi-resolution peak-valley detection execution: (1) Window 3 detection: For the data in window 3, use Neighborhood distance, traversing data to identify peak values ​​(satisfying) ) and valley (satisfying) ), filtering "distance between adjacent extreme points" )” false positives; (2) Similarly, adopt the following methods respectively , The neighborhood distance is used to detect extreme points, and window 1 additionally filters out tiny extreme points with "amplitude difference < 0.2 × N" (to avoid noise interference). (3) Record the sampling index, time scale, and amplitude of each extreme point to form a list of extreme points in three types of windows.

[0039] Step 4: Calculate the frequency, amplitude, and noise level; The noise level update includes performing a noise calculation method, updating the N value, and storing the 5 most recent noise records for every 10 new PMU data points. For the extreme point lists of the three types of windows, perform the following calculation methods respectively: If the number of extreme points is ≥4 groups (window 3 ≥ 2 groups), calculate the frequency f, amplitude A, and confidence level C; If the number of extreme points is insufficient, mark the calculation for this window as invalid. (3) Result filtering: For the valid calculation results of the three types of windows, the group with the highest confidence level is selected (denoted as ). ).

[0040] Step 5: Oscillation Determination and Result Output are as follows: If all criteria are met, a low-frequency oscillation is determined to exist; the oscillation frequency is recorded. Amplitude Confidence level Time marker; If any one of the conditions is not met: it is determined that there is no oscillation, and the oscillation parameter is reset to 0; The detection results will be output in the following manner: The system displays the "oscillation status, frequency, amplitude, and confidence level" in real time. When oscillation occurs, it triggers an audible and visual alarm and uploads the detection results to the power grid dispatch master station at a rate of 25 frames per second using the power system dynamic monitoring system data transmission protocol (DL / T860.92). Historical storage: Stores the detection results of the most recent 72 hours (including raw PMU data segments during the oscillation period, duration = oscillation duration + 2 seconds) for post-event analysis.

[0041] Step 6: Abnormal operating condition adaptation handling is as follows: If the PMU packet loss rate is ≤10%, linear interpolation is used to complete the lost packet data; if the packet loss rate is >10%, detection is suspended and a data anomaly alarm is sent to the main station. When the detected frequency change rate is >0.5Hz / second (e.g., a sudden change from 1.0Hz to 1.6Hz), it is determined to be a frequency change, and the system switches to window 3 priority mode to improve response speed. If active power and voltage phase angle signals are acquired simultaneously, a weighted average is used to fuse the detection results of the two types of signals (power signal weight 0.6, phase angle signal weight 0.4) to further reduce the error.

[0042] Example 3 is an embodiment of the present invention, which provides a low-frequency oscillation identification method based on PMU data, including: The data acquisition module is used to acquire PMU data from the synchronous phasor measurement device through the communication interface, perform packet loss detection and time stamp correction on the data, filter out invalid frame data, and obtain a valid PMU data sequence. The data buffer module is used to establish a global data buffer to store the effective PMU data, set a multi-resolution analysis window covering different oscillation frequency bands according to the PMU sampling rate, and extract the latest data of window 1, window 2 and window 3 from the buffer. The extreme value detection module is used to perform peak and valley value detection on the data in each analysis window, identify extreme points according to the minimum neighborhood distance parameter set according to each window type, and generate the corresponding extreme point sequence. The parameter calculation module is used to calculate the oscillation frequency and oscillation amplitude based on the extreme point sequence, calculate the differential signal of the effective PMU data to obtain the noise level, and adjust the oscillation judgment threshold according to the noise level. The status output module is used to determine the oscillation status based on the oscillation frequency, oscillation amplitude and noise level obtained from each analysis window, and output the detection results.

[0043] Specifically, multi-window differentiated initialization covers full-frequency oscillation. A multi-resolution parallel window detection architecture is adopted, simultaneously maintaining three adaptive windows: Window 1, Window 2, and Window 3, each optimized for the oscillation characteristics of different frequency bands. Window 1 uses an 8-oscillation cycle length to ensure detection accuracy, Window 3 uses a 3-oscillation cycle length to improve response speed, and Window 2 uses a fixed 5-second duration to balance performance requirements.

[0044] Based on PMU sampling rate Minimum detection frequency Maximum detection frequency Construct three types of functional windows, while maintaining a global data buffer for the "maximum window length": Window 1: Length = It covers 8 low-frequency cycles, ensuring the frequency resolution of 0.1~0.5Hz oscillation (error <0.03Hz). Window 2: Length = It covers 5 seconds of data, balancing the accuracy and real-time performance of (0.5~2.0Hz) oscillations (response delay <1 second, error <0.1Hz). Window 3: Length = It covers three high-frequency cycles, ensuring that the response delay of the oscillation (2.0~2.5Hz) is <0.6 seconds and the error is <0.15Hz.

[0045] An adaptive frequency calculation algorithm based on the time interval between extreme points is proposed, employing differentiated frequency calculation methods according to different window types. Window 1 uses multi-period averaging to improve accuracy, window 3 uses few-period calculation to ensure real-time performance, and window 2 uses a balancing strategy. For the signal fluctuation characteristics of the three types of windows in this embodiment, a differentiated "minimum neighborhood distance" (denoted as D, unit: sampling points) is used to detect extreme points. Window 1: It is adapted to the slow-changing characteristics of low-frequency signals.

[0046] Window 3: It is adapted to the rapid fluctuation characteristics of high-frequency signals.

[0047] Window 2: It compromises on neighborhood distance, balancing accuracy and robustness.

[0048] Adaptive frequency-amplitude calculation, matching window characteristics, dynamically selects calculation strategies based on window type, and improves parameter estimation accuracy.

[0049] Window 1 uses the average of the most recent four periods, that is, the average of the time intervals between the most recent four extreme points, denoted as . ,frequency Window 3 uses the "average of the most recent two periods", while window 2 uses the "average of the entire period". Amplitude calculation: based primarily on peak value. Valley minimum value Calculate the amplitude If there are fewer than 3 sets of extreme points, peak-to-peak value calculation using window data is used to ensure that the amplitude error is less than 5%.

[0050] A dynamic noise level estimation mechanism is designed, which calculates the standard deviation of the differential signal through a sliding window, updates the noise level estimate in real time, and dynamically adjusts the detection threshold based on the signal-to-noise ratio to improve detection reliability in noisy environments. This dynamic noise level estimation adapts to complex operating conditions.

[0051] Take the 100 most recent sampling points in the buffer (denoted as ). ), calculate differential signal Noise level (0.8 is a safety factor to avoid misjudgment caused by sudden noise changes), ensuring that the noise estimation lag is <0.1 seconds (hours).

[0052] To achieve deep adaptation with PMU data characteristics, time synchronization analysis is performed using precise time-stamped data provided by the PMU. Criteria are constructed by combining frequency validity, confidence level, signal-to-noise ratio, and amplitude proportion to determine whether oscillations exist. Frequency validity is determined as follows: ; The confidence level is calculated based on the number of effective periods (e.g., confidence level = 1.0 when the number of periods in window 1 is ≥ 4, and confidence level = 0.5 when the number of periods is 2), and must satisfy a confidence level ≥ 0.3; Signal-to-noise ratio: amplitude (Ensure the signal exceeds the noise level by more than 2 times); The percentage of amplitude is ,in The average value of the window data, i.e., the fundamental amplitude, is used to avoid misinterpretation due to minor fluctuations. PMU data deep adaptation ensures time-stamp synchronization and real-time performance, guaranteeing processing efficiency under large data volumes and meeting the real-time monitoring requirements of power systems.

[0053] Extract the time stamp of PMU data to ensure that the timestamps of all sampling points and extreme points are consistent, and the detection results of different PMU measurement points can be compared horizontally. Meanwhile, adopting a point-by-point update + window sliding mode, after new data is passed into the PMU, only the buffer is updated and the extreme points and parameters of the current window are recalculated, with a single-step processing time of less than 1 millisecond.

[0054] The design employs a dual verification mechanism based on amplitude ratio and signal-to-noise ratio, setting minimum amplitude ratio thresholds (0.5%) and minimum signal-to-noise ratio thresholds (2.0) to effectively distinguish between real oscillations and noise interference, thereby reducing the false alarm rate.

[0055] It provides complete detection status output, including oscillation flag, frequency value, amplitude value and confidence level, supports data interaction with the superior monitoring system, and complies with international standards and specifications such as IEC61850.

[0056] This embodiment also provides an electronic device applicable to a low-frequency oscillation identification method based on PMU data, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the low-frequency oscillation identification method based on PMU data as proposed in the above embodiment.

[0057] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a low-frequency oscillation identification method based on PMU data as proposed in the above embodiment.

[0058] The storage medium proposed in this embodiment and the low-frequency oscillation identification method based on PMU data proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0059] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0060] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying low-frequency oscillations based on PMU data, characterized in that: include, PMU data is acquired from the synchronous phasor measurement device through the communication interface, and packet loss detection and time stamp correction are performed on the data to filter out invalid frame data and obtain a valid PMU data sequence. A global data buffer is established to store the valid PMU data; a multi-resolution analysis window covering different oscillation frequency bands is set according to the PMU sampling rate, and the latest data of window 1, window 2, and window 3 are extracted from the buffer; Peak and valley detection is performed on the data within each analysis window. The minimum neighborhood distance parameter is set according to each window type to identify extreme points and generate the corresponding extreme point sequence. The oscillation frequency and oscillation amplitude are calculated based on the extreme point sequence, a differential signal is calculated on the effective PMU data to obtain the noise level, and the oscillation judgment threshold is adjusted according to the noise level. The oscillation state is determined based on the oscillation frequency, oscillation amplitude, and noise level obtained from each analysis window, and the detection results are output.

2. The low-frequency oscillation identification method based on PMU data as described in claim 1, characterized in that: The multi-resolution analysis window includes window 1, window 2, and window 3; Window 1 corresponds to the detection frequency of oscillation signals from 0.1Hz to 0.5Hz, window 2 corresponds to the detection frequency of oscillation signals from 0.5Hz to 2.0Hz, and window 3 corresponds to the detection frequency of oscillation signals from 2.0Hz to 2.5Hz. The length of each window is determined based on the preset number of cycles and sampling rate, forming detection windows for oscillation signals in different frequency bands.

3. The low-frequency oscillation identification method based on PMU data as described in claim 2, characterized in that: The multi-resolution analysis window also includes, When performing extreme point detection, the minimum neighborhood distance parameter is set for different analysis windows. The neighborhood distance used for window 1. The neighborhood distance used for window 2. The neighborhood distance used for window 3; PMU sampling rate, Minimum detection frequency, This is the maximum detection frequency; Window 1 uses neighborhood distance ; Window 2 uses neighborhood distance ; Window 3 uses neighborhood distance ; During the extreme value detection process, false detection points with adjacent spacing less than the neighborhood distance are eliminated based on the neighborhood distance.

4. The low-frequency oscillation identification method based on PMU data as described in claim 3, characterized in that: The oscillation frequency is calculated based on the time interval between adjacent extreme points; Window 1 calculates the frequency using the average time interval of the most recent four cycles; Window 2 uses the average time interval over the entire period to calculate the frequency; Window 3 calculates the frequency using the average time interval of the two most recent cycles.

5. The low-frequency oscillation identification method based on PMU data as described in claim 4, characterized in that: The oscillation amplitude is based on the maximum peak value at the extreme point. and minimum valley value Calculate the amplitude ; When the number of extreme points is less than three sets, the peak value of the window data is used for amplitude calculation.

6. The low-frequency oscillation identification method based on PMU data as described in claim 5, characterized in that: Calculate the differential signal between adjacent sampling points for the valid PMU data. ,in For the current PMU data point, This refers to the previous PMU data point; Calculate the noise level based on the average value of the differential signal. And update the oscillation detection threshold based on the noise level, wherein It is a function of average value.

7. The low-frequency oscillation identification method based on PMU data as described in claim 6, characterized in that: The oscillation state determination is based on a combination of the validity of the oscillation frequency, confidence level, signal-to-noise ratio, and amplitude proportion. When satisfied When the confidence level is greater than 0.3, the signal-to-noise ratio is greater than 2.0, and the amplitude percentage is not less than 0.5%, an oscillation state is determined to exist, and the oscillation frequency, oscillation amplitude, and confidence level are output.

8. A system for identifying low-frequency oscillations based on PMU data, employing the low-frequency oscillation identification method based on PMU data as described in any one of claims 1 to 7, characterized in that, include: The data acquisition module is used to acquire PMU data from the synchronous phasor measurement device through the communication interface, perform packet loss detection and time stamp correction on the data, filter out invalid frame data, and obtain a valid PMU data sequence. The data buffer module is used to establish a global data buffer to store the effective PMU data, set a multi-resolution analysis window covering different oscillation frequency bands according to the PMU sampling rate, and extract the latest data of window 1, window 2 and window 3 from the buffer. The extreme value detection module is used to perform peak and valley value detection on the data in each analysis window, identify extreme points according to the minimum neighborhood distance parameter set according to each window type, and generate the corresponding extreme point sequence. The parameter calculation module is used to calculate the oscillation frequency and oscillation amplitude based on the extreme point sequence, calculate the differential signal of the effective PMU data to obtain the noise level, and adjust the oscillation judgment threshold according to the noise level. The status output module is used to determine the oscillation status based on the oscillation frequency, oscillation amplitude and noise level obtained from each analysis window, and output the detection results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the low-frequency oscillation identification method based on PMU data according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the low-frequency oscillation identification method based on PMU data according to any one of claims 1 to 7.

Images