Power grid disturbance source positioning method and device

Abstract

The application relates to the technical field of power grid quality monitoring, and discloses a power grid disturbance source positioning method and device. The method comprises the following steps: acquiring power grid operation data of a power distribution network through monitoring terminals arranged at multiple monitoring points of the power distribution network; detecting whether the power distribution network has a power grid disturbance based on the power grid operation data; if yes, determining a disturbance source and a corresponding disturbance candidate area based on voltage sag distribution in the power grid operation data; acquiring N-PE voltage waveform data monitored by each monitoring terminal in the disturbance candidate area, and verifying the power grid disturbance based on the N-PE voltage waveform data to obtain a verification result; and positioning the disturbance source in the disturbance candidate area based on frequency wave arrival time differences monitored by each monitoring terminal, geographical position information of each monitoring terminal and the verification result, and determining geographical position information of the disturbance source. The application solves the technical problem that the positioning accuracy is low and unreliable when the disturbance source of the power distribution network is positioned in the related art.

Description

Technical Field

[0001] This application relates to the field of power grid quality monitoring technology, and more specifically, to a method and apparatus for locating power grid disturbance sources. Background Technology

[0002] With the large-scale integration of distributed photovoltaic and wind power and the widespread use of high-power equipment on the user side, disturbance events at the end of the distribution network occur frequently, posing a threat to the safe operation of the power grid. Especially in the high-voltage power grid environment of megacities, the dense substations and extremely low line impedance result in voltage attenuation characteristics during disturbance propagation that differ significantly from those of traditional distribution networks. Related disturbance source localization technologies have the following shortcomings:

[0003] (1) Limitations of a single criterion: Relying solely on voltage amplitude to determine the location of disturbances is susceptible to interference from changes in line impedance and topology switching, resulting in poor positioning accuracy; relying solely on frequency measurement to identify the source of fluctuations involves a huge amount of computation in complex networks and is prone to generating multi-valued solutions due to harmonic interference.

[0004] (2) Dependence on topology information: Most relevant positioning methods rely on the real-time update of the power grid topology. When the power grid undergoes switching or line reconfiguration, the positioning model needs to be adjusted synchronously, otherwise a large error will occur. This places high demands on the real-time synchronization of the background geographic information system and increases the system operation and maintenance costs.

[0005] (3) Data silo problem: Most of the relevant monitoring equipment operates independently, and the collected data has not been effectively aggregated and integrated. The massive amount of terminal data lacks spatiotemporal correlation analysis, which makes it impossible to make judgments based on a single dimension to take advantage of the complementary advantages of multi-source information, which seriously restricts the robustness of positioning.

[0006] (4) Challenges in capturing weak signals in strong power grid environments: In the strong power grid environment of megacities, due to the extremely low loop impedance, the amplitude of the traditional phase voltage (LN) sag is extremely weak, resulting in a severe decrease in the trigger sensitivity and positioning signal-to-noise ratio based on voltage amplitude. At the same time, how to accurately extract the weak characteristic waveforms generated by high-energy remote disturbance coupling from the local strong noise generated by the local load (such as motor starting and inverter switching noise) leads to low disturbance source tracing accuracy.

[0007] There is currently no effective solution to the aforementioned technical problems. Summary of the Invention

[0008] This application provides a method and apparatus for locating power grid disturbance sources, which at least solves the technical problem of low positioning accuracy and unreliability when locating disturbance sources in distribution networks in related technologies.

[0009] According to one aspect of the embodiments of this application, a method for locating a power grid disturbance source is provided, comprising: acquiring power grid operation data of the distribution network through monitoring terminals at multiple monitoring points in the distribution network; detecting whether a power grid disturbance exists in the distribution network based on the power grid operation data; if so, determining the disturbance source and the corresponding disturbance candidate region based on the voltage sag distribution in the power grid operation data; acquiring N-PE voltage waveform data monitored by each monitoring terminal in the disturbance candidate region, and verifying the power grid disturbance based on the N-PE voltage waveform data to obtain a verification result; locating the disturbance source in the disturbance candidate region based on the frequency wave arrival time difference monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification result to determine the geographical location information of the disturbance source.

[0010] Optionally, the step of acquiring power grid operation data of the distribution network through monitoring terminals set at multiple monitoring points in the distribution network includes: acquiring first power grid operation data within a first preset time range before and after the current acquisition time through a dynamic time window; after detecting a power grid disturbance in the distribution network, the method further includes: determining the trigger time of the power grid disturbance; acquiring second power grid operation data within a second preset time range before the trigger time and third power grid operation data within a third preset time range after the trigger time from the first power grid operation data; extracting disturbance features from the second power grid operation data and the third power grid operation data to construct a disturbance fingerprint package, wherein the disturbance fingerprint package includes at least one of the following: monitoring terminal identifier, geographical location information of the monitoring terminal, the trigger time, phase voltage waveform, voltage sag depth, frequency fluctuation curve, frequency change rate, and N-PE ripple characteristics.

[0011] Optionally, determining the disturbance source and corresponding disturbance candidate region based on the voltage sag distribution in the power grid operation data includes: establishing a potential surface gradient model based on the voltage sag depth monitored by each monitoring terminal; sorting the voltage sag depth monitored by each monitoring terminal according to the magnitude based on the potential surface gradient model to obtain the spatial distribution of each voltage sag depth; determining the disturbance source based on the spatial distribution of each voltage sag depth, wherein the disturbance source includes the upstream power grid and local branches of the power grid; and performing preliminary positioning of the disturbance source to determine the disturbance candidate region.

[0012] Optionally, the step of verifying the power grid disturbance based on the N-PE voltage waveform data to obtain a verification result includes: constructing a correlation matrix based on the cross-correlation coefficients between pairs of N-PE voltage waveform data in the N-PE voltage waveform data of each monitoring terminal; determining the average cross-correlation coefficient of each N-PE voltage waveform data based on the correlation matrix, verifying the disturbance source, and obtaining the verification result; wherein, if the average cross-correlation coefficient is greater than a preset correlation coefficient threshold, the disturbance source is determined to be the upstream power grid; the electromechanical wave propagation speed of the power grid disturbance is corrected based on the attenuation gradient of the N-PE voltage waveform data to obtain a first electromechanical wave propagation speed; if the average cross-correlation coefficient is less than or equal to the preset correlation coefficient threshold, the disturbance source is determined to be a local branch of the power grid.

[0013] Optionally, the step of locating the disturbance source within the disturbance candidate area based on the frequency wave arrival time difference monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification result, and determining the geographical location information of the disturbance source, includes: if the frequency change rate of the frequency data in the power grid operation data exceeds a preset frequency change rate threshold, determining that a frequency mutation exists; extracting the start time of the frequency mutation detected by each monitoring terminal within the disturbance candidate area; determining the monitoring terminal with the largest voltage sag depth as the reference monitoring terminal; determining the frequency wave arrival time difference of other monitoring terminals relative to the reference monitoring terminal based on the start time of the frequency mutation of each monitoring terminal and the reference monitoring terminal; constructing a frequency wave arrival time difference equation based on the geographical location information of the monitoring terminals, the frequency wave arrival time difference, and the propagation speed of the electromechanical wave; and solving the frequency wave arrival time difference equation to obtain the geographical location information of the disturbance source.

[0014] Optionally, the above method further includes: correcting the propagation speed of the electromechanical wave using an adaptive inertia correction formula to obtain a second electromechanical wave propagation speed; wherein, the adaptive inertia correction formula is:

[0015] ;

[0016] In the formula, For the propagation speed of electromechanical waves, For equivalent maximum transmission power, The rated angular frequency, For the system inertia density, These are parameters related to the impedance per unit length of the line.

[0017] Optionally, after determining the geographical location information of the disturbance source, the method further includes: identifying the disturbance source type based on the disturbance stiffness index of the disturbance source, wherein the disturbance stiffness index is the coupling ratio of voltage sag depth and frequency deviation, wherein if the disturbance stiffness index is greater than a first preset disturbance stiffness index threshold, the disturbance source type is determined to be a voltage impulse disturbance; if the disturbance stiffness index is less than a second preset disturbance stiffness index threshold, the disturbance source type is determined to be a frequency impulse disturbance, wherein the first preset disturbance stiffness index threshold is greater than the second preset disturbance stiffness index threshold; and / or performing load inertia estimation on the disturbance source based on the disturbance stiffness index and the frequency change rate corresponding to the frequency data in the power grid operation data to determine the system inertia level of the power grid; and correcting the disturbance stiffness index based on the system inertia level of the power grid.

[0018] Optionally, after determining the geographical location information of the disturbance source, the method further includes: extracting N-PE multidimensional features from the N-PE voltage waveform data; and performing feature matching based on the N-PE multidimensional features to identify the disturbance source type.

[0019] Optionally, the method further includes: acquiring the power grid topology information of the distribution network; constructing a priori graph based on the power grid topology information; and correcting the geographical location information of the disturbance source based on the priori graph and a graph variational autoencoder.

[0020] According to another aspect of the embodiments of this application, a power grid disturbance source location device is also provided, comprising: an acquisition module configured to acquire power grid operation data of the distribution network through monitoring terminals at multiple monitoring points in the distribution network; a detection module configured to detect whether there is a power grid disturbance in the distribution network based on the power grid operation data; a determination module configured to determine the disturbance source and the corresponding disturbance candidate region based on the voltage sag distribution in the power grid operation data if there is a power grid disturbance in the distribution network; a verification module configured to acquire N-PE voltage waveform data monitored by each monitoring terminal in the disturbance candidate region, and verify the power grid disturbance based on the N-PE voltage waveform data to obtain a verification result; and a location module configured to locate the disturbance source in the disturbance candidate region based on the frequency wave arrival time difference monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification result to determine the geographical location information of the disturbance source.

[0021] According to another aspect of the embodiments of this application, a non-volatile storage medium is also provided, the non-volatile storage medium including a stored program, wherein, when the program is running, it controls the device where the non-volatile storage medium is located to execute any of the above-described power grid disturbance source location methods.

[0022] According to another aspect of the embodiments of this application, a computer device is also provided, the computer device including a processor, the processor being configured to run a program, wherein the program executes any of the above-described methods for locating power grid disturbance sources.

[0023] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements any of the above-described methods for locating power grid disturbance sources.

[0024] The power grid disturbance source localization method and apparatus provided in this application acquire power grid operation data using dynamic time windows through monitoring terminals at multiple monitoring points in the distribution network. Based on the voltage sag distribution in the power grid operation data, a potential surface gradient model is established to determine the disturbance source and the corresponding disturbance candidate region. This clearly defines the approximate spatial range of the disturbance occurrence, avoiding computational redundancy caused by global search. Furthermore, based on the zero-ground (N-PE) voltage waveform data monitored by each monitoring terminal within the disturbance candidate region, distributed collaborative verification is performed by calculating cross-correlation coefficients to determine the nature of the disturbance source and correct the electromechanical wave propagation speed. This can isolate high-energy long-range disturbance characteristics from local noise, solving the problems of low sensitivity in capturing weak signals and low confidence in determining disturbance sources in strong power grid environments. Finally, based on the arrival time of frequency waves monitored by each monitoring terminal… Based on the geographical location information and verification results of each monitoring terminal, combined with the electromechanical wave propagation speed after inertia adaptive correction, a frequency wave arrival time difference equation is constructed and geometrically solved. This solution process utilizes a unified time base to eliminate time synchronization errors, uses geometric spatial constraints to replace complex power grid topology calculations, and uses multi-node overdetermined equations to eliminate mathematical multivaluedness, thus solving the problem of solution deviation caused by inconsistent time bases, strong topology dependence, and multivalued positioning. Finally, the geographical location information of the disturbance source is obtained based on the solution of the frequency wave arrival time difference equation, efficiently determining the accurate coordinates of the disturbance source from the perspective of spatiotemporal fusion, avoiding large positioning errors caused by the limitations of a single electrical quantity criterion and poor adaptability to dynamic environments. This solves the technical problem of low positioning accuracy and unreliability in the location of disturbance sources in distribution networks in related technologies. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0026] Figure 1 A hardware structure block diagram of a computer terminal for implementing a method for locating power grid disturbance sources is shown.

[0027] Figure 2This is a flowchart illustrating the power grid disturbance source location method provided in the embodiments of this application;

[0028] Figure 3 This is an architecture diagram of a power grid disturbance source location system provided according to an embodiment of this application;

[0029] Figure 4 This is a structural block diagram of a power grid disturbance source locating device provided according to an embodiment of this application. Detailed Implementation

[0030] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0031] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0032] According to an embodiment of this application, a method for locating power grid disturbance sources is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0033] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal for implementing a method for locating power grid disturbance sources is shown. Figure 1As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0034] It should be noted that the aforementioned one or more processors or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits can be implemented wholly or partially as software, hardware, firmware, or any other combination. Furthermore, the data processing circuits can be a single, independent processing module, or wholly or partially integrated into any other element in the computer terminal 10. As involved in the embodiments of this application, the data processing circuit serves as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0035] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the power grid disturbance source location method in this embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby implementing the power grid disturbance source location method of the aforementioned application. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0036] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0037] Figure 2 This is a flowchart illustrating the power grid disturbance source location method provided in the embodiments of this application, as shown below. Figure 2 As shown, the method includes the following steps:

[0038] Step S201: Obtain power grid operation data of the distribution network through monitoring terminals set at multiple monitoring points in the distribution network.

[0039] In this step, multiple monitoring terminals are deployed at different physical locations in the distribution network, and each monitoring terminal collects the power grid operation data of the distribution network in real time.

[0040] Among them, power grid operation data is a collection of electrical quantities reflecting the real-time operating status of the distribution network, including three-phase voltage waveforms, frequency data, voltage sag depth, and zero-to-ground voltage waveform. Specifically, power grid operation data can be obtained through voltage quality recorders.

[0041] This method is applied to a power grid disturbance source location system (hereinafter referred to as the system), which includes a monitoring terminal and a processor. Figure 3 This is an architecture diagram of a power grid disturbance source location system provided according to an embodiment of this application. (For example...) Figure 3 As shown, the monitoring terminal is the sensing layer of the system. The monitoring terminal can be multiple plug-and-play voltage quality recorders deployed in different physical locations on the power grid to collect parameters such as three-phase voltage, frequency, and zero-ground (N-PE) voltage in real time.

[0042] The processor forms the application layer of the system. For example, the processor could be a processor in a cloud processing platform. The perception layer and the application layer are connected through the transport layer. The transport layer includes wireless networks such as WiFi, 4G / 5G, or WAPI secure communication.

[0043] The application layer includes a spatial initial screening module, a zero-ground common-mode ripple discrimination engine, a time-based precision calculation module, a disturbance stiffness analysis module, and a graph neural network correction module. The application layer is responsible for the correlation, fusion, and calculation of multi-point features, ultimately locating the disturbance source and outputting visual results, including the disturbance source's location, type, and confidence level.

[0044] Specifically, such as Figure 3 As shown, the monitoring terminals (such as terminal 1, terminal 2 and terminal 3) upload the collected power grid operation data to the application layer through the transmission layer.

[0045] Based on the above processing, power grid operation data can be obtained and used for subsequent disturbance detection, feature extraction, and location calculation.

[0046] Existing conventional distributed data acquisition devices use local crystal oscillators for timing, lacking a unified high-precision reference time. This makes effective analysis based on waveform arrival time differences impossible, hindering the achievement of cross-node disturbance propagation timing correlation. As an alternative implementation, addressing the issue of inconsistent time bases in existing distributed monitoring terminals leading to limited positioning accuracy, the power grid disturbance source positioning system further integrates a high-precision time synchronization unit. This enables monitoring terminals to achieve network-wide synchronization capabilities, unifying the network's time reference. This unit includes a BeiDou / GPS dual-mode receiver module, a chip-scale atomic clock (CSAC), and a joint disciplined logic circuit. The BeiDou / GPS dual-mode receiver module receives satellite navigation signals and outputs a second pulse signal, providing a global absolute time reference for the entire system. Compared to traditional crystal oscillators, the CSAC, with its extremely low frequency drift characteristics, achieves extremely low time and frequency measurement errors, significantly improving accuracy and ensuring the accuracy of high-frequency feature extraction such as the rate of change of frequency (RoCoF). The joint disciplined logic circuit synchronizes the CSAC output frequency with the satellite reference, further eliminating errors. This capability ensures the reliability of the time reference for locating disturbance sources in complex environments such as industrial plants and basements.

[0047] Specifically, during system operation, multiple plug-and-play monitoring terminals deployed at different physical locations in the power distribution network complete BeiDou / GPS time synchronization through a high-precision time synchronization unit and preset geographical location coordinates so that the disturbance source can be located subsequently based on the time difference of arrival of frequency waves.

[0048] Step S202: Detect whether there is grid disturbance in the distribution network based on grid operation data.

[0049] In this step, based on the power grid operation data obtained in step S201, it is determined whether the effective value of voltage or the rate of change of frequency exceeds a preset threshold, thereby detecting whether there is a power grid disturbance in the distribution network.

[0050] Among them, grid disturbances refer to abnormal fluctuations in voltage, frequency, or phase in the distribution network caused by sudden load changes, faults, or external shocks, including abnormal fluctuations in voltage, frequency, or phase caused by voltage dips, frequency changes, or zero-to-ground voltage anomalies.

[0051] Specifically, such as Figure 3 As shown, the spatial screening module in the application layer determines whether there is a power grid disturbance by using power grid operation data.

[0052] Based on the above processing, it can be determined whether there is a power grid disturbance in the distribution network, and the result of this determination determines whether to trigger the subsequent disturbance source location process.

[0053] Step S203: If yes, determine the disturbance source and the corresponding disturbance candidate region based on the voltage sag distribution in the power grid operation data.

[0054] In this step, when it is determined that there is a power grid disturbance in the distribution network, the source of the disturbance (such as the upstream power grid or local branch) is initially determined based on the voltage sag distribution in the power grid operation data, and the candidate disturbance area is delineated.

[0055] Among them, voltage sag distribution refers to the distribution characteristics of voltage sag depth monitored by each monitoring terminal in the distribution network space, including the voltage sag depth recorded by each terminal and its corresponding spatial location.

[0056] The disturbance candidate region is a geographical area containing potential disturbance sources, initially screened based on the voltage sag distribution. For example, a nearby polygonal or circular area centered on the monitoring terminal with the largest voltage sag depth can be identified as a disturbance candidate region. The voltage sag depth (ΔU) is the decrease in the effective voltage value recorded by the monitoring terminal relative to the nominal voltage, and can be expressed as the difference or proportion between the lowest voltage value during the voltage sag period and the nominal voltage value.

[0057] Specifically, such as Figure 3 As shown, the spatial screening module sorts each monitoring terminal according to the magnitude of the voltage sag depth. The larger the sag magnitude, the closer it is to the disturbance source. Based on the voltage sag distribution, it preliminarily determines whether the disturbance source belongs to the local branch or the upstream grid. It also delineates the disturbance candidate area to narrow down the calculation range for precise positioning.

[0058] Based on the above processing, candidate disturbance regions can be obtained and used for subsequent disturbance verification and localization.

[0059] Step S204: Obtain N-PE voltage waveform data monitored by each monitoring terminal within the disturbance candidate area, and verify the power grid disturbance based on the N-PE voltage waveform data to obtain the verification result.

[0060] In this step, the N-PE voltage waveform data monitored by each monitoring terminal in the disturbance candidate area are obtained, and the source of the power grid disturbance in the distribution network is determined by the spatial correlation between the N-PE voltage waveform data of each terminal, thereby obtaining the verification result.

[0061] Among them, the N-PE voltage waveform data is a dynamic waveform record of the neutral-to-ground voltage, including the waveform characteristics of the neutral-to-ground voltage changing over time as recorded by the monitoring terminal. It is used to analyze the spatial correlation of disturbance sources and thus distinguish the sources of disturbance.

[0062] The verification results are used to characterize the type of disturbance source, including determining whether the disturbance source comes from the upstream power grid or the local branch based on the correlation of N-PE waveform. This information is used to assist in correcting the propagation speed parameters required for subsequent positioning and to guide the selection of subsequent positioning strategies.

[0063] Specifically, such as Figure 3 As shown, the zero-ground common-mode ripple discrimination engine in the application layer verifies grid disturbances based on N-PE voltage waveform data and its spatial correlation.

[0064] Based on the above processing, the verification results can be obtained and used for subsequent high-precision positioning.

[0065] Step S205: Within the candidate disturbance area, the disturbance source is located based on the time difference of arrival of the frequency waves monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification results, thereby determining the geographical location information of the disturbance source.

[0066] In this step, within the candidate disturbance area, the time difference of arrival of frequency waves monitored by each monitoring terminal is extracted. Combining the geographical location information of each monitoring terminal with the above-obtained verification results, the time difference of arrival equation of frequency waves is constructed and solved to determine the geographical location information of the disturbance source.

[0067] The Time Difference of Arrival (TDOA) is the time interval between the arrival of frequency fluctuation signals from the same disturbance source at different monitoring terminals, including the time offset of the start time of each terminal detecting the power grid disturbance relative to the reference terminal.

[0068] The geographical location information of each monitoring terminal is the physical location coordinates of each monitoring terminal in the Earth coordinate system, including longitude, latitude coordinates or relative coordinates relative to a reference point.

[0069] The geographical location information of the disturbance source is the spatial location of the disturbance event point that has been calculated and finally determined, including the coordinates of the tower, transformer or line section where the disturbance source is located.

[0070] Specifically, such as Figure 3 As shown, after initially defining the candidate disturbance region and completing the N-PE verification, the system starts the time precision calculation module to perform precise positioning based on the TDOA algorithm.

[0071] Based on the above processing, the geographical location information of the disturbance source can be obtained, and the mapping from electrical quantity measurement to geospatial location is finally realized.

[0072] Through the steps S201-S205 described above, the presence of grid disturbances in the distribution network is detected based on grid operation data, enabling real-time perception of abnormal grid conditions. Then, based on the voltage sag distribution in the grid operation data, the disturbance source and corresponding candidate disturbance region are determined. The spatial distribution characteristics of the voltage sag depth are used to initially pinpoint the approximate disturbance range, reducing the computational complexity of the global search. Next, N-PE voltage waveform data monitored by each monitoring terminal within the candidate disturbance region are acquired, and the grid disturbance is verified based on this data. The verification results are obtained, and the spatial correlation characteristics of the zero-ground voltage waveform are used to assist in determining the nature of the disturbance, enhancing anti-interference capabilities and positioning reliability in strong grid environments. Finally, within the candidate disturbance region, the disturbance source is located based on the frequency wave arrival time difference monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification results, determining the geographical location information of the disturbance source. This efficiently determines the optimal disturbance source location from the feasible set, avoiding low decision-making efficiency and unreliable positioning due to low accuracy of a single criterion or inconsistent time bases. This solves the technical problem of low positioning accuracy and unreliability in related technologies when locating disturbance sources in distribution networks.

[0073] As an optional embodiment, in step S201, power grid operation data of the power distribution network is acquired through monitoring terminals set at multiple monitoring points in the power distribution network, including:

[0074] S2011, acquire the first power grid operation data within the first preset time range before and after the current acquisition time through a dynamic time window;

[0075] S2012, After detecting a power grid disturbance in the distribution network, determine the trigger time of the power grid disturbance;

[0076] S2013, Obtain second power grid operation data within a second preset time range before the trigger time and third power grid operation data within a third preset time range after the trigger time from the first power grid operation data;

[0077] S2014. Disturbance features are extracted from the second power grid operation data and the third power grid operation data to construct a disturbance fingerprint package. The disturbance fingerprint package includes at least one of the following: monitoring terminal identifier, monitoring terminal geographical location information, trigger time, phase voltage waveform, voltage sag depth, frequency fluctuation curve, frequency change rate, and N-PE ripple features.

[0078] In steps S2011 to S2014 above, a dynamic time window recording mechanism is introduced on the monitoring terminal side as the core hardware support for improving positioning accuracy.

[0079] The dynamic time window refers to setting a buffer to continuously and cyclically store the latest collected data, covering and retaining continuous sampling data within the most recent period (i.e., the first preset time range).

[0080] The trigger time is the current acquisition time when the monitoring terminal detects that the power grid disturbance exceeds the limit and locks the starting point of the disturbance, including the timestamp of the moment the disturbance occurs based on the high-precision time synchronization signal.

[0081] The first power grid operation data consists of continuous electrical quantity data recorded within a dynamic time window, covering a preset time range before and after the current acquisition time. This includes historical steady-state waveforms before the trigger time and transient waveforms after the trigger time, which are the second and third power grid operation data. The second power grid operation data includes steady-state voltage and frequency waveforms within a period before the disturbance occurs (i.e., the second preset time range); the third power grid operation data includes transient voltage, frequency, and N-PE voltage waveform data within a period after the disturbance occurs (i.e., the third preset time range).

[0082] Disturbance characteristics are key electrical parameters extracted from power grid operation data that characterize the nature of power grid disturbances, including phase voltage waveforms, voltage sag depth, frequency fluctuation curves, frequency change rate, N-PE ripple characteristics, etc.

[0083] The disturbance fingerprint packet is a data structure that carries key information and metadata about the disturbance event. It includes a combination data packet containing the monitoring terminal identifier and its geographical location information, the trigger time, and the disturbance characteristics.

[0084] Specifically, such as Figure 3 As shown, a dynamic time window mechanism is introduced on the terminal side of the sensing layer. By setting up a ring buffer in the monitoring terminal, the power grid operation data (such as three-phase voltage waveform, frequency data and N-PE voltage waveform data) of the most recent period (such as 400ms) are continuously collected and stored. The sampling rate of the N-PE channel is not less than 6.4kHz.

[0085] When the monitoring terminal detects a voltage dip (i.e., the effective voltage value drops beyond a preset threshold) or an excessive rate of frequency change, it immediately locks the current acquisition time and determines the trigger time of the disturbance source. This trigger time is based on BeiDou / GPS time and can achieve microsecond-level accuracy.

[0086] Then, the transient waveform data from a period of time before the trigger time (e.g., 200ms) to the historical steady-state waveform data from a period of time after the trigger time (e.g., 200ms) are read from the ring buffer. At the same time, disturbance features are further extracted, including voltage sag waveforms, frequency change rate curves, and N-PE ripple features.

[0087] Finally, the generated disturbance fingerprint packet is uploaded to the application layer, including: monitoring terminal ID, geographical location information, trigger time, waveforms before and after triggering, voltage sag depth, frequency fluctuation curve, RoCoF, N-PE ripple characteristics, etc.

[0088] It is understandable that frequency fluctuation signals (electromechanical waves) typically follow voltage dips, with a slight time difference (from several milliseconds to tens of milliseconds) between them. By saving waveform data prior to the trigger time, the cloud processing platform can accurately reconstruct the absolute time of the electromechanical waves arriving at each monitoring terminal, which is the foundation for achieving precise TDOA positioning. Simultaneously, the retrospective recording of N-PE voltage waveform data provides complete time-domain information for the zero-ground common-mode ripple discrimination engine in the application layer.

[0089] Based on the above processing, by introducing a dynamic time window mechanism, the positioning accuracy is improved from the kilometer level to the hundred-meter level.

[0090] As an optional embodiment, step S203, determining the disturbance source and corresponding disturbance candidate region based on the voltage sag distribution in the power grid operation data, includes:

[0091] S2031, Based on the voltage sag depth monitored by each monitoring terminal, establish a potential surface gradient model;

[0092] S2032, based on the potential surface gradient model, sorts the voltage sag depths monitored by each monitoring terminal according to their magnitude to obtain the spatial distribution of voltage sag depths;

[0093] S2033, Based on the spatial distribution of voltage sag depth, determine the disturbance source, wherein the disturbance source includes the upstream power grid and the local branch of the power grid;

[0094] S2034, perform preliminary location of disturbance source and determine disturbance candidate region.

[0095] In steps S2031 to S2034 above, the spatial preliminary screening module determines the disturbance source and the corresponding disturbance candidate region by establishing a potential surface gradient model.

[0096] The potential surface gradient model is a mathematical model that maps the voltage sag depth monitored by each monitoring terminal onto a spatial geometric surface. It includes a spatial surface distribution model constructed with the geographical location of the monitoring terminal as coordinates and the voltage sag depth as the potential value. The spatial distribution of the voltage sag depth refers to the arrangement and distribution of the voltage sag depth in the geographical space of the distribution network, including the gradient distribution features formed on the map with the magnitude of the voltage sag depth as the weight.

[0097] Specifically, such as Figure 3As shown, the spatial preliminary screening module aggregates the voltage sampling drop depths (i.e., voltage sag depths) reported by each monitoring terminal and establishes a potential surface gradient model. Then, the monitoring terminals are sorted according to the magnitude of the voltage sag depths to obtain the spatial distribution of the voltage sag depths. Furthermore, based on this spatial distribution, it is preliminarily determined whether the disturbance source belongs to the upstream power grid or a local branch of the power grid. Finally, the corresponding candidate disturbance regions are delineated to roughly locate the disturbance source, narrowing down the calculation range for precise location.

[0098] The upstream power grid is the external power supply network that supplies power to this distribution network, i.e., the transmission network, including substation busbars or high-voltage transmission networks. Local branches of the power grid are local power supply units or line segments within the distribution network, including a specific transformer area or branch feeder.

[0099] The larger the magnitude of the voltage sag depth (the deeper the sag depth, the smaller the residual voltage), the closer the monitoring terminal is to the disturbance source. Faults in the upstream power grid can cause voltage sags over a wide area (hundreds of kilometers), while the impact of short-circuit faults in the local distribution network is smaller; moreover, voltage sag depths caused by transmission network faults are shallower, while those caused by local distribution network faults are deeper.

[0100] For example, when the voltage sag depths monitored by each monitoring terminal are all small (the number of monitoring terminals with voltage sag depths less than a preset depth threshold determined by the ranking results is greater than a first preset threshold), and the location range of each monitoring terminal is large (for example, the multiple monitoring terminals with small voltage sag depths are far apart from each other), the disturbance source is determined to be the upstream power grid, and the transmission line corresponding to a certain number of monitoring terminals with the largest voltage sag depth amplitude is determined as a disturbance candidate area according to the amplitude ranking results. When the voltage sag depths monitored by each monitoring terminal are large (the number of monitoring terminals with voltage sag depths greater than a preset depth threshold determined by the ranking results is greater than a second preset threshold), and the distance between each monitoring terminal is relatively short, the disturbance source is determined to be a local branch of the power grid, and the branch line or transformer area corresponding to the monitoring terminal with the largest voltage sag depth amplitude is determined as a disturbance candidate area.

[0101] Based on the above processing, by establishing a potential surface gradient model, abstract electrical quantities can be transformed into intuitive spatial distribution characteristics. By sorting and analyzing spatial distribution patterns, it is possible to effectively distinguish whether disturbances originate from the upper-level power grid or local branches, thereby quickly determining the disturbance source type and delineating preliminary candidate areas, providing a prerequisite for subsequent high-precision positioning.

[0102] In the above embodiments, the disturbance source is initially located based on the difference in voltage sag depth and amplitude and the location information of the monitoring terminal. In specific implementations, the initial location can also be based on the duration of the voltage sag. For example, voltage sags caused by transmission network faults have a short duration, about 100ms; voltage sags caused by local distribution network faults have a longer duration than those caused by transmission network faults.

[0103] As an optional embodiment, in step S204, the grid disturbance is verified based on the N-PE voltage waveform data to obtain the verification result, including:

[0104] S2041, Based on the cross-correlation coefficients between pairs of N-PE voltage waveform data in the N-PE voltage waveform data of each monitoring terminal, construct a correlation matrix;

[0105] S2042, based on the correlation matrix, determine the average cross-correlation coefficient of each N-PE voltage waveform data, verify the disturbance source, and obtain the verification result;

[0106] S2043, If the average cross-correlation coefficient is greater than the preset correlation coefficient threshold, the disturbance source is determined to be the upstream power grid;

[0107] S2044, based on the attenuation gradient of N-PE voltage waveform data, correct the electromechanical wave propagation speed of grid disturbance to obtain the first electromechanical wave propagation speed;

[0108] S2045, if the average cross-correlation coefficient is less than or equal to the preset correlation coefficient threshold, the disturbance source is determined to be a local branch of the power grid.

[0109] In S2041 to S2045 above, the zero-ground (N-PE) common-mode ripple discrimination engine constructs a correlation matrix based on N-PE voltage waveform data, and then performs attribution verification and wave velocity correction for grid disturbances.

[0110] The cross-correlation coefficient is a statistical measure of the similarity between N-PE voltage waveform data collected by two different monitoring terminals in the time domain. It includes a measure of the correlation between one waveform and another waveform under time shift. The higher the cross-correlation coefficient, the more similar the waveforms are.

[0111] The correlation matrix is ​​a matrix with each monitoring terminal as the row and column index and the cross-correlation coefficient as the element. It includes a two-dimensional data table describing the waveform similarity relationship between all monitoring terminals.

[0112] The average cross-correlation coefficient is the average of all off-diagonal elements or specific sub-matrix elements in the correlation matrix, representing the overall similarity of N-PE voltage waveforms among all monitoring terminals in the candidate region;

[0113] The preset correlation coefficient threshold is a benchmark value used to distinguish different types of disturbance sources, including a numerical limit set between 0 and 1. When the average cross-correlation coefficient is higher than the preset correlation coefficient threshold, it indicates that the N-PE voltage waveform is highly synchronized, and the verification result is determined to be a disturbance coupled from the upstream power grid; when it is lower than the preset correlation coefficient threshold, it indicates that the waveform difference is large, and the verification result is determined to be a disturbance from the local branch.

[0114] The attenuation gradient is the rate at which the amplitude or characteristic intensity of the N-PE voltage waveform changes with the increase of the electrical distance between the monitoring terminal and the disturbance source, including the decreasing trend of the similarity or amplitude of the N-PE voltage waveform with increasing distance.

[0115] The first electromechanical wave propagation speed is the electromechanical wave propagation speed after attenuation gradient correction, which reflects the macroscopic propagation characteristics of electromechanical waves on a specific path, but does not take into account the dynamic physical parameters inside the power grid.

[0116] Specifically, such as Figure 3 As shown, after determining the candidate disturbance region, i.e. after the spatial initial screening module and before the time fine calculation module, the system calls the N-PE voltage waveform data synchronously recorded by all monitoring terminals within the candidate disturbance region, starts the zero-ground common-mode ripple discrimination engine, and performs the following collaborative verification:

[0117] (1) Calculate the cross-correlation coefficients between each pair of N-PE voltage waveforms at each monitoring terminal to form a correlation matrix;

[0118] (2) If the average cross-correlation coefficient between monitoring terminals within the disturbance candidate area is greater than the preset threshold (e.g.) If the value is 0.85, then the disturbance source is determined to be high-voltage hierarchical coupling injection (i.e., the disturbance comes from the upstream power grid and exhibits high spatial consistency in a strong power grid environment).

[0119] (3) Use the attenuation gradient of N-PE voltage waveform data to assist in coarse directional compensation and correct the initial estimate of electromechanical wave propagation speed;

[0120] (4) If the average cross-correlation coefficient is less than or equal to the preset threshold (e.g.) If the value is 0.5, the disturbance is determined to be generated within the local branch. In this case, no complex wave velocity correction is needed, and the positioning range can be directly narrowed down to a single unit area or branch.

[0121] Based on the above processing, the reliability of disturbance source attribution determination is improved by spatial correlation analysis of N-PE voltage waveform data; and the problem of insufficient sensitivity caused by relying solely on phase voltage amplitude in strong power grid environments is effectively solved by the collaborative verification mechanism.

[0122] As an optional embodiment, in step S205, within the disturbance candidate area, the disturbance source is located based on the time difference of arrival of the frequency waves monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification results, and the geographical location information of the disturbance source is determined, including:

[0123] S2051, If ​​the frequency change rate of the frequency data in the power grid operation data exceeds the preset frequency change rate threshold, it is determined that there is a frequency mutation; the start time of the frequency mutation detected by each monitoring terminal in the disturbance candidate area is extracted;

[0124] S2052, determine the monitoring terminal with the largest voltage sag depth as the reference monitoring terminal;

[0125] S2053, Based on the start time of the frequency change of each monitoring terminal and the reference monitoring terminal, determine the frequency wave arrival time difference of other monitoring terminals relative to the reference monitoring terminal;

[0126] S2054, based on the geographical location information of the monitoring terminal, the time difference of arrival of frequency waves and the propagation speed of electromechanical waves, a time difference of arrival equation for frequency waves is constructed;

[0127] S2055, solve the frequency wave arrival time difference equation to obtain the geographical location information of the disturbance source.

[0128] In steps S2051 to S2055 above, the time calculation module, in conjunction with the reference monitoring terminal, determines the arrival time difference of frequency waves from other monitoring terminals, and then obtains the geographical location information of the disturbance source by constructing and solving equations.

[0129] The rate of change of frequency (RoCoF) is the rate at which the power grid frequency changes over time, including the absolute value or differential change of frequency per unit time.

[0130] The preset frequency change rate threshold is a benchmark value used to determine whether a significant frequency disturbance has occurred, including a pre-set upper or lower limit for the frequency change rate.

[0131] The reference monitoring terminal is the monitoring terminal with the largest voltage sag depth in the candidate area, that is, the monitoring terminal that is closest to the disturbance source or the one most affected.

[0132] The frequency wave arrival time difference equation is a set of mathematical equations describing the geometric relationship between the coordinates of the disturbance source, the coordinates of the monitoring point, the time difference, and the wave velocity. It describes a hyperbolic relationship with each monitoring terminal as the focal point and the distance difference corresponding to the time difference as the focal length. By solving multiple equations in the set of equations simultaneously, the two-dimensional or three-dimensional coordinates of the disturbance source can be determined.

[0133] Specifically, after initially delineating the candidate disturbance region and completing the zero-ground (N-PE) collaborative verification, the system activates the time precision calculation module to perform precise positioning based on the time difference of arrival (TDOA) of electromechanical waves:

[0134] (1) Extract the absolute recording time of frequency mutations (RoCoF exceeding the preset threshold) detected by each monitoring terminal within the disturbance candidate area. Since the monitoring terminal has the ability to backtrack before triggering, it can accurately capture the starting moment when the frequency wave begins to deviate from the steady-state value;

[0135] (2) Using the monitoring terminal with the largest voltage sag depth as the reference point, calculate the frequency wave arrival time difference of other monitoring terminals relative to the reference point;

[0136] (3) The time difference positioning algorithm is adopted, combined with the electromechanical wave propagation speed (first electromechanical wave propagation speed) after being corrected by zero ground (N-PE) collaborative verification. A set of TDOA equations was established. This set of equations only involves the geographical coordinates of the monitoring terminal and the time difference of arrival (TDOA) of the frequency waves, and does not involve the electrical connections of the power grid.

[0137] .

[0138] (4) By solving the overdetermined equations, the geographic coordinates (latitude and longitude) of the disturbance source can be directly locked, and the positioning error can be controlled within the range of 100 meters.

[0139] Based on the above processing, without relying on the power grid topology, the geographical location information of the disturbance source is accurately determined by geometric calculation based solely on the time characteristics of electromechanical wave propagation, thus achieving topology-independent high-precision positioning.

[0140] As an optional embodiment, the power grid disturbance source localization method further includes: correcting the electromechanical wave propagation speed using an adaptive inertia correction formula to obtain a second electromechanical wave propagation speed; wherein, the adaptive inertia correction formula is:

[0141] ;

[0142] In the formula, For the propagation speed of electromechanical waves, For equivalent maximum transmission power, The rated angular frequency, For the system inertia density, These are parameters related to the impedance per unit length of the line.

[0143] In this embodiment, the time calculation module corrects the electromechanical wave propagation speed using an adaptive inertia correction formula to obtain a second electromechanical wave propagation speed. Here, the electromechanical wave propagation speed before correction using the adaptive inertia correction formula includes the first electromechanical wave speed (determined as a fault in the upper-level power grid, obtained using the attenuation gradient) and the electromechanical wave propagation speed without attenuation gradient correction (determined as a fault in the local branch, requiring no attenuation gradient correction).

[0144] The inertia adaptive correction formula is a mathematical expression describing the relationship between the propagation speed of electromechanical waves and the electrical parameters of the power grid and the system inertia. This formula is based on a wave speed estimation model constructed using the power grid's transmission power, rated frequency, system inertia density, and line impedance parameters. The equivalent maximum transmission power is a parameter characterizing the synchronization capability or rigidity between power grid nodes. This equivalent maximum transmission power is a power limit value that is proportional to the square of the inter-node voltage and inversely proportional to the line reactance. The rated angular frequency is the angular velocity corresponding to the nominal frequency of the power grid, including the fixed angular frequency value corresponding to a 50Hz or 60Hz system. The system inertia density is an inertia parameter reflecting the power grid's resistance to frequency changes, including the inertia time constant provided by rotating machinery or equivalent energy storage devices in the system. The line impedance-related parameters are impedance coefficients describing the electrical characteristics of the line, including the reactance or impedance modulus per unit length of line.

[0145] Specifically, in the above adaptive inertia correction formula, the propagation speed of the electromechanical wave... It is not a constant; its value depends on two dynamic variables: the system's inertia density and the equivalent maximum transmission power, resulting in two effects.

[0146] (1) Mass effect: speed of propagation of electromechanical waves Inversely proportional to inertia ( System inertia density This is equivalent to the mass of a power system. Under the same disturbance impact force, the greater the system inertia, the slower the frequency change, and the more "heavy" and slow the propagation of electromechanical waves.

[0147] The manifestation of this mass effect in weak inertia systems is the fundamental reason for the dynamic correction of electromechanical wave propagation speed in this application. That is, with the large-scale integration of new energy sources such as photovoltaics, the system inertia density... Decreasing the denominator in the formula reduces the propagation speed of electromechanical waves. Significantly faster. Without dynamic correction, directly applying traditional empirical values ​​to calculate the arrival time difference of frequency waves... This will result in a significant deviation in the positioning distance.

[0148] (2) Spring effect: speed of propagation of electromechanical waves Proportional to synchronous power ( ).because Characterizing the rigidity or synchronization capability of a power grid, this is analogous to the "spring" stiffness between connecting nodes. The more compact the power grid architecture (the smaller the line reactance), the better. The larger the value, the stiffer the "spring," and the faster the disturbance signal is transmitted. Therefore, when the power grid topology changes (such as switching or line reconfiguration), the equivalent maximum transmission power... This change, in turn, affects the propagation speed of electromechanical waves. The above formula provides a theoretical basis for wave speed correction based on inertia adaptation.

[0149] Based on the above processing, the equivalent maximum transmission power, rated angular frequency, system inertia density, and line impedance parameters of the power grid are incorporated into the calculation of electromechanical wave propagation speed using the inertia adaptive correction formula. This can dynamically reflect the changes in the characteristics of the power grid under operating conditions such as switch switching, load changes, or new energy access, thereby obtaining a more accurate electromechanical wave propagation speed and improving the reliability of the time difference-based positioning algorithm.

[0150] As an optional embodiment, after determining the geographical location information of the disturbance source, the power grid disturbance source location method further includes:

[0151] S301, based on the disturbance stiffness index of the disturbance source, identifies the disturbance source type, where the disturbance stiffness index is the coupling ratio of voltage sag depth to frequency deviation;

[0152] S302, if the disturbance stiffness index is greater than the first preset disturbance stiffness index threshold, determine the disturbance source type as voltage impulse disturbance;

[0153] S303, if the disturbance stiffness index is less than the second preset disturbance stiffness index threshold, the disturbance source type is determined to be frequency impulse disturbance, and the first preset disturbance stiffness index threshold is greater than the second preset disturbance stiffness index threshold.

[0154] In steps S301 to S303 above, the disturbance stiffness analysis module distinguishes disturbance source types based on the disturbance stiffness index and by setting different threshold ranges. Simultaneously, it estimates the load inertia by combining the frequency change rate and corrects the disturbance stiffness index.

[0155] The disturbance stiffness index is a comprehensive indicator that quantifies the impact characteristics of disturbance sources on the power grid. This index reflects the relative strength of voltage changes with respect to frequency changes during a disturbance event and is a key physical parameter used to distinguish the types of disturbance sources.

[0156] The first preset disturbance stiffness index threshold is the upper limit for judging voltage impulse disturbance. For example, when the disturbance stiffness index exceeds this threshold, it is judged as a voltage impulse disturbance.

[0157] The second preset disturbance stiffness index threshold is the lower limit for determining frequency impact disturbances. For example, when the disturbance stiffness index is lower than this threshold, it is determined to be a frequency impact disturbance, and this threshold is less than the first preset disturbance stiffness index threshold.

[0158] Specifically, such as Figure 3 As shown, after locating the disturbance source, the system initiates the disturbance stiffness analysis module to identify the type of disturbance source. The Disturbance Stiffness Index (DSI) is defined as the voltage sag depth. Frequency deviation The coupling ratio is calculated using the following formula:

[0159] ;

[0160] in, Nominal voltage, This is the nominal frequency.

[0161] It is understandable that DSI reflects the impact stiffness of the disturbance source on the power grid. The larger the DSI, the more likely the disturbance is a voltage surge (purely resistive load or near-end fault); the smaller the DSI, the more likely the disturbance is a frequency surge (large inertia equipment or power deficit).

[0162] As an optional embodiment, the power grid disturbance source location method further includes:

[0163] S304. Based on the disturbance stiffness index and the frequency change rate corresponding to the frequency data in the power grid operation data, the load inertia of the disturbance source is estimated to determine the system inertia level of the power grid.

[0164] S305, based on the system inertia level of the power grid, corrects the disturbance stiffness index.

[0165] Load inertia estimation is a process of estimating the inertia level of the power grid or the system near the disturbance point based on the disturbance stiffness index and the frequency change rate. It is used to quantify the power grid's resistance to frequency disturbances and help assess the weak inertia state of the system.

[0166] System inertia level is an indicator characterizing the magnitude of the overall or local inertia of a power grid, reflecting the total inertia of rotating machinery in the system. High inertia means slower frequency changes, while low inertia means faster frequency changes.

[0167] The disturbance stiffness analysis module can estimate the system inertia level of the disturbance source through joint analysis of DSI and RoCoF:

[0168] ;

[0169] in, Disturbance power refers to the change in active power that causes disturbances in the power grid, such as the power surge value when high-power equipment starts up or disconnects from the grid. This represents the maximum rate of frequency change after the disturbance occurs.

[0170] Then, the disturbance stiffness analysis module normalizes the DSI index based on the system inertia level. Specifically, (1) the inertia correction coefficient is calculated. ,in, The system inertia is usually the system inertia under rated operating conditions. (2) The calculated original disturbance stiffness index Multiplying by the inertia correction factor yields the corrected disturbance stiffness exponent. Alternatively, based on the system's inertia density. The DSI threshold range for determining the type of disturbance is dynamically adjusted.

[0171] Through the above corrections or dynamic threshold adjustments, the impact of changes in power grid operating conditions (such as inertia fluctuations) on the disturbance type identification results is eliminated, ensuring the stability and accuracy of disturbance source type identification under different operating conditions.

[0172] Based on the above processing, the introduction of a disturbance stiffness index enables the identification of disturbance source types beyond a single electrical quantity and allows for the quantitative assessment of the impact characteristics of disturbance sources on the power grid. By setting different threshold ranges, different types of disturbance sources can be clearly distinguished. Simultaneously, load inertia is estimated based on the disturbance stiffness index and the frequency change rate, and the disturbance stiffness index is then corrected. This eliminates the bias in disturbance characteristic identification under different system inertia levels, ensuring the stability and accuracy of disturbance source type identification under different operating conditions when the power grid operating status changes (e.g., fluctuations in new energy output leading to changes in system inertia).

[0173] In one alternative implementation, the system inertia density Real-time estimation is achieved through the following steps:

[0174] Using small-amplitude frequency fluctuation data and corresponding system active power disturbance data collected during normal grid operation, and based on a linearized model of the swing equation, the system inertia level at the current moment is estimated using recursive least squares method or sliding window regression analysis. Among them, small frequency fluctuation data refers to normal disturbance data with a rate of change of frequency (RoCoF) less than a preset safety threshold, in order to eliminate nonlinear response interference caused by major faults.

[0175] In one optional implementation, the trigger threshold of the monitoring terminal is dynamically adjusted based on the real-time estimated system inertia level, specifically including:

[0176] Establish trigger threshold and system inertia density The mapping function is given, where the trigger threshold is positively correlated with the system inertia density; when the system inertia density... When the voltage sag trigger threshold decreases, the algorithm automatically increases the voltage sag trigger threshold and the rate of change of frequency (RoCoF) trigger threshold; conversely, when the system inertia density decreases... When the threshold increases, the algorithm automatically lowers the trigger threshold.

[0177] Based on the above processing, the monitoring terminal maintains high sensitivity in high inertia systems, while suppressing false triggering caused by frequent fluctuations in low inertia (high noise, easy fluctuation) systems.

[0178] As an optional embodiment, after determining the geographical location information of the disturbance source, the power grid disturbance source location method further includes: extracting N-PE multidimensional features from N-PE voltage waveform data; and performing feature matching based on the N-PE multidimensional features to identify the disturbance source type.

[0179] In this embodiment, the zero-ground common-mode ripple discrimination engine extracts N-PE multidimensional features from the N-PE voltage waveform data and performs matching to identify the disturbance source type.

[0180] Among them, the N-PE multidimensional features are feature vectors of multiple dimensions extracted from N-PE voltage waveform data to characterize disturbance characteristics, including at least one of time-domain features (such as peak value, RMS value), frequency-domain features (such as dominant frequency, harmonic content), and time-frequency-domain features (such as wavelet coefficients, entropy value).

[0181] Feature matching is the process of comparing or classifying the extracted N-PE multidimensional features with standard features in a pre-stored disturbance feature library or database. For example, it involves calculating feature distance, using classification algorithms or template matching techniques to determine the most similar disturbance pattern, thereby identifying the disturbance source type. This disturbance source type includes large motor starting, capacitor switching, photovoltaic inverter disconnection from the grid, grounding faults, or external grid impacts.

[0182] It is understandable that different types of disturbances exhibit different characteristics in the N-PE dimension. When subjected to external power surges, the N-PE ripple exhibits high-frequency, constant-amplitude oscillations, and the synchronization rate of multiple monitoring terminals reaches an extremely high correlation with microsecond-level precision. When a local large motor starts, the N-PE voltage only shows a significant DC bias rise in the branch circuit and its nearest terminal, and the spatial correlation decreases rapidly with electrical distance. When a remote grounding fault occurs, the N-PE voltage exhibits asymmetrical drift, and the distribution transformer area to which the fault point belongs can be verified by comparing the phase offset difference of the N-PE voltage monitored by multiple monitoring terminals. When a purely resistive load is switched on or off, the N-PE ripple has no obvious characteristics and the response is transient. When a short-circuit fault occurs on the distribution network side, the N-PE voltage waveform is strongly coupled with the phase voltage (LN), and the synchronization of multiple points is high. When a photovoltaic inverter is disconnected from the grid, the N-PE voltage may exhibit high-frequency oscillations, and the spatial correlation is relatively weak.

[0183] Based on the above processing, feature matching is performed using N-PE multidimensional features to identify the type of disturbance source. This allows for refined classification of disturbance properties by utilizing the unique high-frequency oscillations, DC bias, or phase shift fingerprints generated on the zero-ground channel by different disturbance sources (such as photovoltaic inverter grid disconnection, large motor startup, and grounding faults). Through this multidimensional feature fingerprint matching, the system can not only locate the disturbance position but also accurately identify its physical cause, overcoming the limitations of relying solely on voltage and frequency amplitude characteristics to distinguish complex disturbance types. This provides richer decision-making basis for targeted operation and maintenance and fault isolation of the power grid.

[0184] As an optional embodiment, the power grid disturbance source localization method further includes: acquiring the power grid topology information of the distribution network; constructing a priori graph based on the power grid topology information; and correcting the geographical location information of the disturbance source based on the priori graph and the graph variational autoencoder.

[0185] In this embodiment, the graph neural network correction module uses a graph variational autoencoder, with multidimensional feature vectors as node features and the power grid prior topology as the graph structure, to learn the disturbance propagation law, thereby outputting the accurate positioning coordinates after correcting the preliminary positioning coordinates of the disturbance source.

[0186] The power grid topology information typically originates from a Geographic Information System (GIS) ledger. This ledger records the attribute parameters and electrical connections of components such as buses, lines, transformers, and switches in the distribution network. It is structured data describing the physical structure of the power grid and the connections between electrical nodes. A priori graph constructed based on the power grid topology information is used to represent the potential connections between power grid nodes. For example, this priori graph includes a graph structure with monitoring terminals as vertices and connections in the power grid ledger as edges, serving as the input basis for subsequent deep learning models.

[0187] Graph Variational Autoencoders (VAEs) are deep learning models based on graph neural networks that learn low-dimensional latent representations of graph data and reconstruct graph structures or node features. For example, a Graph Variational Autoencoder uses the initial estimate of TDOA geometric localization as a spatial constraint, multi-dimensional physical features such as DSI as node attributes, and a pre-extended graph as a structural basis to learn the propagation patterns of disturbances in electrical networks, thereby uncovering the implicit relationships between topology and disturbance features.

[0188] Understandably, in order to further improve positioning accuracy, especially when some power grid topology information is available (e.g., the accuracy of GIS ledgers is about 90%), this application introduces a graph variational autoencoder (Graph VAE) for post-processing correction based on the initial estimate of TDOA geometric positioning, so as to achieve dual-mode fusion positioning of "physical model + data-driven".

[0189] Specifically, such as Figure 3 As shown, the graph neural network correction module receives the preliminary location coordinates of the disturbance source based on the time difference of arrival (TDOA) of electromechanical waves, and performs spatial coordinate correction by combining the multi-dimensional feature vectors of each monitoring terminal. These multi-dimensional feature vectors include: voltage sag depth, frequency change rate, disturbance stiffness index (DSI), N-PE voltage ripple characteristics, and the geographical coordinates of the monitoring terminal. The working process of the graph neural network correction module is as follows:

[0190] (1) Construct the initial graph structure. Each monitoring terminal is used as a node in the graph network. For each monitoring terminal node, a multi-dimensional feature vector is extracted to form a node feature matrix. The node features include: the initial estimated coordinates of the disturbance source obtained based on the TDOA algorithm, voltage sag depth, frequency change rate RoCoF, N-PE zero-ground voltage ripple characteristics, geographic coordinate information of the monitoring terminal, and disturbance stiffness index (DSI).

[0191] (2) Construct the connection relationships between nodes, i.e., edges. Optionally, construct a prior adjacency matrix based on the static topology information in the power grid GIS ledger. Due to potential update delays or data errors in GIS ledgers, this prior adjacency matrix... It characterizes the potential theoretical connections between power grid nodes with an accuracy of approximately 90%. As an alternative, when power grid topology information is completely missing, it can be based on the geographical distance between monitoring terminals. Construction of K-Nearest Neighbors (KNN) algorithm The nearest neighbor graph serves as the prior edges to provide initial topological assumptions.

[0192] (3) Perform graph variational coding. Use a graph variational encoder to process the input node feature matrix. and prior adjacency matrix The encoder performs processing. Through graph convolution operations, it maps high-dimensional node features and topology to a low-dimensional latent space, learning the implicit distribution relationship between the location of disturbance sources and the power grid graph structure. Specifically, the encoder outputs latent variables. mean and variance Its probability distribution is approximately:

[0193] ;

[0194] in, These are latent variables used to capture the true location characteristics of disturbance sources and the dynamic topology changes of the power grid during disturbances. 3. Perform graph variational decoding. The graph variational decoder is used to extract information from the latent variables. Reconstruct node features and adjacency matrices. The decoder aims to recover observed data and infer potential structural changes; its reconstruction process is represented as follows:

[0195] ;

[0196] in, The reconstructed node features include the corrected coordinates of the disturbance source, which have higher accuracy than the initial TDOA estimate, down to the tower level. The reconstructed adjacency matrix reflects the actual power grid topology inferred from the measurement data, thus enabling adaptive correction of the prior adjacency matrix. Errors in the process (such as incorrect switch states or connection relationships).

[0197] (4) Calculate the loss function and apply physical constraints. During model training, a loss function containing three components is constructed. To optimize latent variables And the reconstruction results.

[0198] During training, the loss function consists of three terms:

[0199] ;

[0200] in, The reconstruction loss is used to force the reconstructed node features (especially the coordinates of the disturbance source) to approximate the actual location of the disturbance source, thereby improving the positioning accuracy. for The divergence term, as a variational regularization term, prevents the model from overfitting and allows the model to tolerate the prior adjacency matrix in the prior graph within a certain range. Topological error; The physical constraint loss includes electrical distance consistency constraints and voltage drop constraints, which are used to ensure that the reconfiguration result conforms to the physical operating laws of the power system.

[0201] (5) Positioning correction process output.

[0202] The initial TDOA coordinates, fingerprint packet features uploaded by each terminal (including N-PE ripple features, DSI, etc.), and prior adjacency matrix are used. The input is fed into the trained graphical variational autoencoder. The model outputs the following results:

[0203] (1) Corrected disturbance source coordinates: accuracy can reach the tower level;

[0204] (2) The corrected graph structure: i.e. the inferred true topological relationships ;

[0205] (3) Confidence level: assess the reliability of the location result.

[0206] Based on the above processing, the graph neural network module can use electrical topology information to compensate for and correct errors caused by geometric positioning, thereby achieving accurate positioning at the tower level in situations where the power grid topology is complex or there is switching.

[0207] The following describes the single-phase line fault determination device provided in this application embodiment using specific examples.

[0208] Example 1: Starting and Positioning of Large Motors in Industrial Park

[0209] Twelve plug-and-play monitoring terminals were deployed in an industrial park, each connected to a three-phase socket in one of the factory buildings. One day, the cloud processing platform received trigger events reported by eight of these terminals. The system executed the following steps:

[0210] 1. Dynamic Time Window Backtracking: After each terminal detects a voltage sag, it extracts voltage waveform data from the 200ms preceding the trigger from the circular buffer. Analysis shows that the frequency fluctuation begins approximately 28ms after the voltage sag, providing an accurate benchmark for calculating the Time Difference of Arrival (TDOA).

[0211] 2. Initial spatial screening: Analysis of voltage sag data revealed that the terminal sag amplitude of plant A was the largest (12%), and it was preliminarily determined that the disturbance source was located near plant A.

[0212] 3. Distributed N-PE Collaborative Verification: N-PE voltage waveform data of each terminal within the candidate area were retrieved, and the cross-correlation coefficient was calculated. Results showed that N-PE DC bias rise only occurred at the terminals of Plant A and its adjacent plants. The spatial correlation decreased rapidly with electrical distance, and the average cross-correlation coefficient... It is determined to be a local branch disturbance, and no remote beam velocity correction is required.

[0213] 4. Time-of-Arrival Calculation (Topology Independent): The starting time of the frequency fluctuation curves of each terminal is extracted, and the time difference of arrival of the frequency wave at each terminal is calculated through cross-correlation analysis. Combining the power grid operating state parameters and the real-time estimated value of inertia (approximately 2.5s), the corrected electromechanical wave propagation speed is approximately 15km / s. The Time Difference of Arrival (TDOA) equation is constructed and solved, and the location of the disturbance source is calculated to be approximately 35 meters west of Plant A. This calculation does not depend on the line connection relationship between plants, but is based solely on geographical coordinates and time difference.

[0214] 5. Analysis of Disturbance Stiffness and N-PE Integration: The calculated DSI = (12% / 220V) / (0.15Hz / 50Hz) ≈ 1.8 matches the "Large Motor Starting" type. The N-PE ripple characteristics show a localized DC bias rise, consistent with the local motor starting characteristics.

[0215] 6. Results Output: The system marks the location of the disturbance source (motor equipment on the west side of Factory A) on the electronic map and pushes an early warning: a high-power motor has been detected starting. Location: west side of Factory A. Affected area: Factory A and three surrounding factories. Disturbance stiffness index 1.8 (medium inertia equipment). N-PE local response characteristics match local start.

[0216] The above optional implementation methods can achieve at least the following beneficial effects:

[0217] (1) This application introduces a collaborative verification mechanism for N-PE voltage waveform data, which utilizes the cross-correlation coefficient and attenuation gradient of the zero-ground voltage waveform in the spatially deployed array to effectively distinguish between disturbances coupled into the upper-level power grid and disturbances generated within the local branch. This mechanism solves the problem of insufficient sensitivity caused by relying solely on phase voltage amplitude in strong power grid environments, and can accurately extract high-energy remote disturbance characteristics from local noise, significantly improving trigger sensitivity and positioning reliability in weak signal scenarios.

[0218] (2) This application constructs an overdetermined set of equations based on the time difference of arrival (TDOA) of frequency waves and the geographical location information of the monitoring terminal for geometric solution, thereby realizing blind source localization that is independent of the power grid topology. Since this localization process only depends on the geographical coordinates of the terminal and the waveform time difference of arrival, and is completely independent of the power grid line topology and switch status, the localization algorithm can work normally without any modification even if the power grid undergoes switch switching, line reconfiguration, or operation mode adjustment. This greatly reduces the requirement for real-time synchronization of the background GIS database and enhances the robustness and adaptability of the system.

[0219] (3) Compared with related technologies, this application has achieved a leap from single location positioning to a comprehensive profile of "location-type-characteristics" by constructing a Disturbance Stiffness Index (DSI) and combining it with N-PE multidimensional features for integrated analysis. The DSI index can quantitatively assess the impact intensity of disturbances on the power grid, distinguish between voltage impact type (such as purely resistive loads and near-end faults) and frequency impact type (such as large inertia equipment and power deficit) disturbances, and further identify specific disturbance source types (such as motor starting and photovoltaic disconnection) by combining N-PE features. This provides operation and maintenance personnel with richer and more accurate decision-making basis than traditional methods, and improves the efficiency of distribution network fault operation and maintenance.

[0220] According to an embodiment of this application, a power grid disturbance source location device for implementing the above-described power grid disturbance source location method is also provided. Figure 4 This is a structural block diagram of a power grid disturbance source locating device provided according to an embodiment of this application, such as... Figure 4 As shown, the power grid disturbance source locating device includes:

[0221] The acquisition module 100 is configured to acquire power grid operation data of the power distribution network through monitoring terminals set at multiple monitoring points in the power distribution network;

[0222] The detection module 200 is configured to detect whether there are power grid disturbances in the distribution network based on power grid operation data;

[0223] The determination module 300 is configured to determine the disturbance source and the corresponding disturbance candidate region based on the voltage sag distribution in the power grid operation data if there is a power grid disturbance in the power distribution network.

[0224] The verification module 400 is configured to acquire N-PE voltage waveform data monitored by each monitoring terminal within the disturbance candidate area, and to verify the power grid disturbance based on the N-PE voltage waveform data to obtain the verification result;

[0225] The positioning module 500 is configured to locate the disturbance source within the disturbance candidate area based on the time difference of arrival of frequency waves monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification results, and to determine the geographical location information of the disturbance source.

[0226] It should be noted that the acquisition module 100, detection module 200, determination module 300, verification module 400, and positioning module 500 mentioned above correspond to steps S201 to S205 in the embodiments. Multiple modules and their corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in the above embodiments. It should also be noted that the above modules, as part of the device, can run on the computer terminal 10 provided in the embodiments.

[0227] Embodiments of this application may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0228] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the power grid disturbance source location method and device in this application embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned power grid disturbance source location method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to a computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0229] The processor can access the information and application programs stored in the memory via the transmission device to execute the steps S201 to S205 described above.

[0230] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0231] Embodiments of this application also provide a non-volatile storage medium. Optionally, in this embodiment, the aforementioned non-volatile storage medium can be used to store the program code executed by the power grid disturbance source localization method provided in the above embodiments.

[0232] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0233] Embodiments of this application also provide a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can implement the methods of steps S201 to S205 described above.

[0234] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0235] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0236] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0237] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0238] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0239] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0240] The above are merely optional embodiments of this application. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for locating sources of power grid disturbance, characterized in that, include: The power grid operation data of the power distribution network is obtained through monitoring terminals set at multiple monitoring points in the power distribution network; Based on the power grid operation data, detect whether there is a power grid disturbance in the distribution network; If so, the disturbance source and the corresponding disturbance candidate region are determined based on the voltage sag distribution in the power grid operation data; Obtain the N-PE voltage waveform data monitored by each monitoring terminal within the disturbance candidate area, and verify the power grid disturbance based on the N-PE voltage waveform data to obtain the verification result; Within the candidate disturbance area, the disturbance source is located based on the time difference of arrival of the frequency waves monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification result, thereby determining the geographical location information of the disturbance source.

2. The method according to claim 1, characterized in that, The acquisition of power grid operation data of the distribution network through monitoring terminals installed at multiple monitoring points in the distribution network includes: The first power grid operation data within the first preset time range before and after the current acquisition time is obtained through a dynamic time window; After detecting a power grid disturbance in the distribution network, the method further includes: Determine the trigger time of the power grid disturbance; Obtain second power grid operation data within a second preset time range before the trigger time and third power grid operation data within a third preset time range after the trigger time from the first power grid operation data; Disturbance features are extracted from the second power grid operation data and the third power grid operation data to construct a disturbance fingerprint package. The disturbance fingerprint package includes at least one of the following: monitoring terminal identifier, geographical location information of the monitoring terminal, trigger time, phase voltage waveform, voltage sag depth, frequency fluctuation curve, frequency change rate, and N-PE ripple features.

3. The method according to claim 1, characterized in that, The process of determining the disturbance source and corresponding disturbance candidate region based on the voltage sag distribution in the power grid operation data includes: Based on the voltage sag depth monitored by each of the monitoring terminals, a potential surface gradient model is established. Based on the potential surface gradient model, the voltage sag depths monitored by each monitoring terminal are sorted according to their amplitude to obtain the spatial distribution of the voltage sag depths. Based on the spatial distribution of the voltage sag depth, the disturbance source is determined, wherein the disturbance source includes the upstream power grid and local branches of the power grid; The disturbance source is initially located to determine the candidate disturbance region.

4. The method according to claim 3, characterized in that, The verification of the power grid disturbance based on the N-PE voltage waveform data, and the resulting verification results, include: Based on the cross-correlation coefficients between any two N-PE voltage waveform data points in the N-PE voltage waveform data of each monitoring terminal, a correlation matrix is ​​constructed. Based on the correlation matrix, the average cross-correlation coefficient of each N-PE voltage waveform data is determined, and the disturbance source is verified to obtain the verification result; wherein, if the average cross-correlation coefficient is greater than a preset correlation coefficient threshold, the disturbance source is determined to be the upstream power grid; based on the attenuation gradient of the N-PE voltage waveform data, the electromechanical wave propagation speed of the power grid disturbance is corrected to obtain the first electromechanical wave propagation speed; If the average cross-correlation coefficient is less than or equal to the preset correlation coefficient threshold, the disturbance source is determined to be a local branch of the power grid.

5. The method according to claim 4, characterized in that, Within the candidate disturbance area, the disturbance source is located based on the time difference of arrival of the frequency waves monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification result, and the geographical location information of the disturbance source is determined, including: If the frequency change rate of the frequency data in the power grid operation data exceeds a preset frequency change rate threshold, a frequency mutation is determined to exist. Extract the start time of the frequency mutation detected by each monitoring terminal within the disturbance candidate region; The monitoring terminal with the largest voltage sag depth is designated as the reference monitoring terminal. Based on the start time of the frequency change of each of the monitoring terminals and the reference monitoring terminal, the arrival time difference of the frequency wave of the other monitoring terminals relative to the reference monitoring terminal is determined; Based on the geographical location information of the monitoring terminal, the time difference of arrival of the frequency wave, and the propagation speed of the electromechanical wave, a time difference of arrival equation for the frequency wave is constructed. Solve the time difference of arrival equation for the frequency wave to obtain the geographical location information of the disturbance source.

6. The method according to claim 5, characterized in that, The method further includes: The propagation speed of the electromechanical wave is corrected by the inertia adaptive correction formula to obtain the second propagation speed of the electromechanical wave. The adaptive inertia correction formula is as follows: ； In the formula, For the propagation speed of electromechanical waves, For equivalent maximum transmission power, The rated angular frequency, For the system inertia density, These are parameters related to the impedance per unit length of the line.

7. The method according to claim 1, characterized in that, After determining the geographical location information of the disturbance source, the method further includes: Based on the disturbance stiffness index of the disturbance source, the disturbance source type is identified, wherein the disturbance stiffness index is the coupling ratio of voltage sag depth to frequency deviation, wherein if the disturbance stiffness index is greater than a first preset disturbance stiffness index threshold, the disturbance source type is determined to be a voltage impulse disturbance; if the disturbance stiffness index is less than a second preset disturbance stiffness index threshold, the disturbance source type is determined to be a frequency impulse disturbance, wherein the first preset disturbance stiffness index threshold is greater than the second preset disturbance stiffness index threshold; and / or Based on the disturbance stiffness index and the frequency change rate corresponding to the frequency data in the power grid operation data, the load inertia of the disturbance source is estimated to determine the system inertia level of the power grid; the disturbance stiffness index is corrected based on the system inertia level of the power grid.

8. The method according to claim 1, characterized in that, After determining the geographical location information of the disturbance source, the method further includes: Extract N-PE multidimensional features from the N-PE voltage waveform data; Based on the N-PE multidimensional features, feature matching is performed to identify the type of disturbance source.

9. The method according to any one of claims 1 to 6, characterized in that, The method further includes: Obtain the power grid topology information of the distribution network; A priori graph is constructed based on the aforementioned power grid topology information; The geographical location information of the disturbance source is corrected based on the prior map and the graph variational autoencoder.

10. A power grid disturbance source locating device, characterized in that, include: The acquisition module is configured to acquire power grid operation data of the power distribution network through monitoring terminals set at multiple monitoring points in the power distribution network; The detection module is configured to detect whether there is a power grid disturbance in the distribution network based on the power grid operation data; The determination module is configured to determine the disturbance source and the corresponding disturbance candidate region based on the voltage sag distribution in the power grid operation data if there is a power grid disturbance in the distribution network. The verification module is configured to acquire N-PE voltage waveform data monitored by each monitoring terminal within the disturbance candidate area, and to verify the power grid disturbance based on the N-PE voltage waveform data to obtain the verification result; The positioning module is configured to locate the disturbance source within the disturbance candidate area based on the time difference of arrival of the frequency waves monitored by each monitoring terminal, the geographical location information of each monitoring terminal, and the verification result, and determine the geographical location information of the disturbance source.

Images