Power distribution network fault locating method and device based on satellite time service inverter

Abstract

The application discloses a power distribution network fault positioning method and device based on satellite timing inverter, belongs to the technical field of power system fault diagnosis, and comprises the following steps: acquiring a timing signal of a satellite navigation system, decoding and generating a unified space-time reference, distributing the unified space-time reference to a plurality of inverter nodes in a power distribution network, and constructing a cooperative measurement network; injecting a coded sequence current signal into a power distribution network line by a master inverter node; synchronously collecting response voltage traveling wave data by each inverter node, and performing cross-correlation analysis on the response voltage traveling wave data and the coded sequence current signal to extract the propagation time delay of each inverter node; and correcting the reference line wave velocity by using current line parameters and current meteorological data, and combining the propagation time delay to calculate fault point position data. By using satellite timing to construct an inverter cooperative measurement network, actively injecting a coded sequence current signal, and combining the means of correcting real-time wave velocity by using an intelligent model, the power distribution network fault can be positioned quickly, accurately and with high reliability.

Description

Technical Field

[0001] This invention relates to the field of power system fault diagnosis technology, and in particular to a method and apparatus for locating distribution network faults based on satellite time-synchronized inverters. Background Technology

[0002] The power distribution network is a crucial component of the power system, and its safe and stable operation directly impacts the quality of electricity supply for users. With the widespread integration of distributed power sources, especially photovoltaic power generation, distributed nodes using inverters as interfaces are becoming increasingly common in distribution networks. When a fault occurs in a distribution network line, quickly and accurately locating the fault point is essential for shortening power outage time and improving power supply reliability. The traveling wave method, unaffected by fault type or transition resistance, is currently recognized as one of the most promising fault location technologies.

[0003] Existing fault location technologies for distribution networks based on the traveling wave method typically involve installing traveling wave acquisition devices at substations or key nodes of the line. These devices capture the traveling wave signal generated at the moment of a fault and utilize its propagation characteristics to calculate the fault distance. The dual-end traveling wave location method is one of the more accurate approaches. It determines the fault location by comparing the time difference between the arrival times of the traveling wave at the measuring devices at both ends of the line. Theoretically, this method can eliminate dependence on the traveling wave velocity, but it requires strict time synchronization between the measuring devices at both ends.

[0004] However, achieving high-precision time synchronization among multiple measuring devices across the entire network is difficult and expensive; even small deviations in time synchronization can directly translate into significant positioning errors. Secondly, the traveling wave signals generated by faults are random and uncertain. Particularly for high-resistance grounding faults, the initial traveling wave amplitude is weak, and the wavefront characteristics are indistinct, making them easily submerged under complex power grid background noise interference, thus making it difficult to accurately extract the arrival time of the traveling wave. Furthermore, the propagation speed of traveling waves in the line is not constant; it is affected by line material, aging, and environmental factors such as temperature and humidity. Traditional methods mostly use fixed theoretical wave velocity values ​​for calculation, which does not match reality and is another significant source of positioning errors. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a method and apparatus for locating distribution network faults based on satellite-synchronized inverters. By utilizing satellite timing to construct an inverter collaborative measurement network, actively injecting coded sequence current signals, and combining this with intelligent model correction of real-time wave velocity, rapid, accurate, and highly reliable fault location in the distribution network can be achieved.

[0006] The above objectives can be achieved through the following approach:

[0007] A method for fault location in a distribution network based on a satellite-synchronized inverter includes: acquiring timing signals from a satellite navigation system, decoding to generate a unified spatiotemporal reference, and distributing the unified spatiotemporal reference to multiple inverter nodes in the distribution network to form a collaborative measurement network; based on the unified spatiotemporal reference, the master control inverter node in the collaborative measurement network injects a coded sequence current signal with sharp single-peak autocorrelation characteristics into the distribution network line; based on the unified spatiotemporal reference, each inverter node in the collaborative measurement network synchronously collects the corresponding response voltage traveling wave data generated by the coded sequence current signal; performing cross-correlation analysis on the coded sequence current signal and the response voltage traveling wave data of each inverter node to extract the propagation time delay of each inverter node; acquiring current line parameters, current meteorological data, and the reference line wave velocity corresponding to the current line, and using the current line parameters and the current meteorological data to correct the reference line wave velocity corresponding to the current line to obtain the real-time wave velocity; and using the propagation time delay and the real-time wave velocity to calculate the fault location data.

[0008] Optionally, the construction of the collaborative measurement network includes: acquiring and decoding the timing signal of the satellite navigation system to obtain time synchronization information and location information; fusing the time synchronization information and the location information to generate a unified spatiotemporal reference; and distributing the unified spatiotemporal reference to each inverter node through power line carrier communication or fiber optic network to construct the collaborative measurement network.

[0009] Optionally, the step of injecting a coded sequence current signal with sharp single-peak autocorrelation characteristics into the distribution network line by the master control inverter node in the collaborative measurement network based on the unified spatiotemporal reference includes: generating a coded sequence fundamental signal with sharp single-peak autocorrelation characteristics based on the unified spatiotemporal reference; modulating the coded sequence fundamental signal to obtain a coded sequence current signal, and injecting it into the distribution network line through the master control inverter node in the collaborative measurement network.

[0010] Optionally, the step of synchronously acquiring the corresponding response voltage traveling wave data generated by the coded sequence current signal by each inverter node in the collaborative measurement network based on the unified spatiotemporal reference includes: configuring a sampling time window for each inverter node based on the unified spatiotemporal reference; acquiring the raw response voltage traveling wave data generated by the coded sequence current signal within the sampling time window; and preprocessing the raw response voltage traveling wave data to remove power frequency components and high-frequency noise, thereby generating response voltage traveling wave data.

[0011] Optionally, the extraction of the propagation time delay of each inverter node includes: performing cross-correlation calculation on the encoded sequence current signal and the response voltage traveling wave data of each inverter node to generate a cross-correlation coefficient sequence curve for each inverter node; identifying the peak position in the cross-correlation coefficient sequence curve of each inverter node to extract the initial time delay of each inverter node; performing consistency verification on the initial time delay of each inverter node, and generating the propagation time delay of each inverter node based on the verification result.

[0012] Optionally, before obtaining the real-time wave velocity, the method further includes: acquiring historical line parameters, historical meteorological data, historical traveling wave propagation speed, and the reference line wave velocity corresponding to the historical line; calculating the difference between the historical traveling wave propagation speed and the reference line wave velocity corresponding to the historical line to obtain the historical wave velocity deviation; and establishing and training a neural network model using the historical line parameters and the historical meteorological data as inputs and the historical wave velocity deviation as outputs to obtain a wave velocity error prediction model.

[0013] Optionally, obtaining the real-time wave velocity includes: acquiring current line parameters, current meteorological data, and the reference line wave velocity corresponding to the current line; inputting the current line parameters and the current meteorological data into the wave velocity error prediction model to obtain the current wave velocity deviation; using the current wave velocity deviation to correct the reference line wave velocity corresponding to the current line to obtain the initial wave velocity; and performing segmented calibration of the initial wave velocity according to the segment differences of the current line to obtain the real-time wave velocity.

[0014] Optionally, the step of calculating the fault location data using the propagation time delay and the real-time wave velocity includes: calculating the time difference of the traveling wave arriving at each inverter node based on the propagation time delay; combining the time difference and the real-time wave velocity to calculate the preliminary location of the fault point through dual-end traveling wave positioning; and using redundant data in the collaborative measurement network to verify and optimize the preliminary location, thereby outputting the fault location data.

[0015] Optionally, the method further includes: recording the propagation time delay, line parameters, and meteorological data of each inverter node at the time of the fault; calculating the actual measured value of the line wave speed during the fault period by combining the fault location data and the propagation time delay of each inverter node at the time of the fault; and correcting the wave speed error prediction model by using the actual measured value, the line parameters, and the meteorological data at the time of the fault.

[0016] Based on the same inventive concept, this invention also provides a power distribution network fault location device based on a satellite timing inverter. The device includes: a timing network construction module, used to acquire timing signals from a satellite navigation system, decode and generate a unified spatiotemporal reference, and distribute the unified spatiotemporal reference to multiple inverter nodes in the power distribution network to form a collaborative measurement network; a signal injection module, used to inject a coded sequence current signal with sharp single-peak autocorrelation characteristics into the power distribution network lines from the master control inverter node in the collaborative measurement network based on the unified spatiotemporal reference; and a response acquisition module, used to receive a response from each inverter node in the collaborative measurement network based on the unified spatiotemporal reference. The inverter nodes synchronously acquire the corresponding response voltage traveling wave data generated by the coded sequence current signal; the correlation analysis module is used to perform cross-correlation analysis between the coded sequence current signal and the response voltage traveling wave data of each inverter node to extract the propagation time delay of each inverter node; the wave speed correction module is used to acquire the current line parameters, current meteorological data and the reference line wave speed corresponding to the current line, and use the current line parameters and the current meteorological data to correct the reference line wave speed corresponding to the current line to obtain the real-time wave speed; the fault location module is used to calculate the fault location data using the propagation time delay and the real-time wave speed.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] 1. This invention utilizes a satellite navigation system to provide a unified spatiotemporal reference for inverters widely distributed in the power distribution network, constructs a high-precision synchronous collaborative measurement network, eliminates time synchronization errors between measurement nodes, lays the foundation for accurate calculation of traveling wave propagation time, and improves the overall accuracy of fault location.

[0019] 2. This invention employs a main control inverter to actively inject a coded sequence current signal with sharp single-peak autocorrelation characteristics into the line, transforming fault location from relying on passive fault signals with uncertain characteristics to analyzing the response of known detection signals. Combined with cross-correlation analysis technology, it enhances the detection capability and wavefront identification accuracy of weak traveling wave signals in strong noise background, thereby improving the anti-interference capability and reliability of the location method under various fault conditions.

[0020] 3. This invention introduces a real-time dynamic wave velocity correction mechanism based on line parameters and meteorological data. By establishing an intelligent prediction model and using historical and real-time data to accurately correct the traveling wave propagation velocity, it overcomes the systematic error caused by the use of fixed wave velocity values ​​in traditional methods. This enables the positioning calculation to adapt to the dynamic changes in the line operating environment, further improving the accuracy and environmental adaptability of the positioning results.

[0021] 4. This invention establishes a closed-loop self-learning optimization mechanism. After each successful fault location, it can use the confirmed fault information to back-calculate the actual line wave speed and use this high-value data to iteratively update the wave speed error prediction model, so that the positioning performance of the device can be continuously improved with the accumulation of operating experience, and it has the ability to perform long-term adaptive optimization.

[0022] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating the distribution network fault location method based on a satellite time-synchronized inverter according to an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of the autocorrelation characteristics of the encoded sequence current signal according to an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram of the wave velocity error prediction model relationship in an embodiment of the present invention.

[0027] Figure 4 This is a schematic diagram of the structure of a power distribution network fault location device based on a satellite time-synchronized inverter according to an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Reference Figure 1One embodiment of the present invention proposes a distribution network fault location method based on satellite time-synchronized inverters. By utilizing satellite time synchronization to construct an inverter collaborative measurement network, actively injecting coded sequence current signals, and combining intelligent model correction of real-time wave velocity, the method can achieve rapid, accurate, and highly reliable fault location in the distribution network.

[0030] The method described in this embodiment specifically includes:

[0031] S1. Obtain the timing signal from the satellite navigation system, decode it to generate a unified spatiotemporal reference, and distribute the unified spatiotemporal reference to multiple inverter nodes in the power distribution network to form a collaborative measurement network.

[0032] Optionally, the constituting the collaborative measurement network includes:

[0033] The timing signal from the satellite navigation system is acquired and decoded to obtain time synchronization information and position information.

[0034] By integrating the time synchronization information and the location information, a unified spatiotemporal reference is generated;

[0035] The unified spatiotemporal reference is distributed to each inverter node via power line carrier communication or fiber optic network to form a collaborative measurement network.

[0036] Specifically, one or more key inverter nodes in the distribution network are equipped with high-precision satellite navigation system receiver modules. These modules continuously receive broadcast signals from satellite navigation systems such as BeiDou and GPS. First, time synchronization information and location information are separated and parsed from the complex satellite signals. Time synchronization information is typically represented by a highly precise pulse-of-seconds (PPS) signal along with an accompanying Coordinated Universal Time (UTC) timestamp. Location information refers to the precise three-dimensional geographic coordinates of the inverter node where the receiver module is located, including longitude, latitude, and altitude. Subsequently, the decoded time synchronization information and location information are fused to generate a unified spatiotemporal reference. This fusion binds a high-precision time standard with precise spatial coordinates, providing a reference framework for all measurement activities in the network. This means that any event in the network can not only be accurately recorded at its moment of occurrence but also associated with its physical location, thus laying the foundation for subsequent calculations of traveling wave propagation time and distance. Finally, to ensure that all inverter nodes in the distribution network share this unified spatiotemporal reference, it must be distributed through a reliable communication channel.

[0037] This method offers two technical approaches. The first is Power Line Carrier Communication (PLC) technology, which modulates unified spatiotemporal reference information onto a carrier signal of a specific frequency, utilizing existing power lines as the transmission medium to distribute it to each inverter node in the network. Its advantage lies in the elimination of the need for additional communication lines, resulting in high cost-effectiveness. The second approach is a fiber optic network, distributing the information through dedicated fiber optic communication links. Fiber optic communication offers advantages such as high bandwidth, low latency, and strong resistance to electromagnetic interference, ensuring the integrity and real-time performance of the unified spatiotemporal reference information during transmission. Once each inverter node receives the unified spatiotemporal reference and calibrates its local clock and position data, they transform from a series of independent devices into a unified whole, forming a collaborative measurement network with high-precision synchronous measurement capabilities, providing the technical prerequisite for subsequent fault location procedures.

[0038] S2. Based on the unified spatiotemporal reference, the master control inverter node in the collaborative measurement network injects a coded sequence current signal with sharp single-peak autocorrelation characteristics into the distribution network line.

[0039] Specifically, this operation is an active probing step performed by the master inverter node designated as having signal injection capability within the network after the collaborative measurement network completes unified spatiotemporal reference synchronization. This operation first generates a fundamental signal of a coded sequence with sharp single-peak autocorrelation characteristics. This fundamental signal is typically a predefined pseudo-random sequence, such as an m-sequence. Autocorrelation is an indicator that measures the similarity between a signal and itself at different time delays. The sharp single-peak characteristic ensures that the sequence exhibits a significant strong correlation peak only when the time delay is zero; at any other delay, its correlation value is close to zero. This forms the physical basis for accurately extracting the signal propagation time from a strong noise background. Subsequently, based on the established unified spatiotemporal reference, the master inverter node, at a predetermined precise time, utilizes its internal power electronic converter and, according to the symbol logic of the fundamental signal of the coded sequence, precisely controls the current waveform output to the distribution network line using high-frequency switching modulation technology, thereby generating and injecting the final coded sequence current signal. The injected coded sequence current signal carries known coded information and propagates on the power line as an artificially applied active detection signal, providing a stable and easily identifiable signal source for subsequent synchronous response acquisition and related analysis of each inverter node.

[0040] S3. Based on the unified spatiotemporal reference, each inverter node in the collaborative measurement network synchronously collects the corresponding response voltage traveling wave data generated by the encoded sequence current signal;

[0041] Optionally, the step of synchronously acquiring the corresponding response voltage traveling wave data generated by the coded sequence current signal based on the unified spatiotemporal reference by each inverter node in the collaborative measurement network includes:

[0042] Based on the unified spatiotemporal reference, a sampling time window is configured for each inverter node;

[0043] Within the sampling time window, raw response voltage traveling wave data generated by the coded sequence current signal are acquired;

[0044] The original response voltage traveling wave data is preprocessed to remove power frequency components and high-frequency noise, thereby generating response voltage traveling wave data.

[0045] Specifically, the device controller calculates and issues a precise sampling time window command to each inverter node in the collaborative measurement network based on a unified spatiotemporal reference. The start time of this sampling time window is usually set with a small time margin before the predetermined time when the master inverter node injects the coded sequence current signal. Its duration is determined based on the maximum electrical distance of the distribution network and the theoretical propagation speed of traveling waves, ensuring that the window is sufficient to fully cover the entire process of the traveling wave propagating from the injection point to any node in the network and possibly returning after being reflected by a fault point.

[0046] Upon receiving the instruction, each inverter node will activate its high-speed data acquisition unit within a strictly synchronized sampling time window under the unified spatiotemporal reference. This unit samples the instantaneous value of the line voltage at a sampling rate on the order of megahertz using high-precision voltage transformers and analog-to-digital converters (ADCs). This high-frequency sampling process captures all voltage dynamics excited by the propagation and reflection of the injected coded sequence current signal in the distribution network, forming raw response voltage traveling wave data containing the desired signal, power frequency components, and various types of noise. After acquisition, each inverter node or central server needs to perform digital signal preprocessing on the obtained raw response voltage traveling wave data. The core purpose of this preprocessing is to highlight the weak traveling wave signal. First, a digital bandpass filter is used to remove the extremely strong power frequency components and randomly generated high-frequency noise. The passband frequency range of this filter is precisely designed to match the spectral characteristics of the previously injected coded sequence current signal, thereby preserving the integrity of the useful signal to the greatest extent while filtering out interference. This preprocessing step can be expressed as:

[0047] ,

[0048] in, For discrete time point indexes, This is the sequence of raw response voltage traveling wave data collected from each inverter node. This represents a digital bandpass filtering operation, which is implemented by a digital signal processor (DSP) or software algorithm, and its parameters are determined by the spectral characteristics of the injected signal. This refers to the preprocessed response voltage traveling wave data that is ultimately generated for subsequent analysis.

[0049] S4. Perform cross-correlation analysis on the encoded sequence current signal and the response voltage traveling wave data of each inverter node to extract the propagation time delay of each inverter node.

[0050] Optionally, the extraction of the propagation time delay of each inverter node includes:

[0051] The encoded sequence current signal is cross-correlated with the response voltage traveling wave data of each inverter node to generate the cross-correlation coefficient sequence curve of each inverter node.

[0052] Identify the peak position in the cross-correlation coefficient sequence curve of each inverter node and extract the initial time delay of each inverter node.

[0053] The initial time delay of each inverter node is checked for consistency, and the propagation time delay of each inverter node is generated based on the check results.

[0054] Specifically, the first step is to perform cross-correlation calculations on the previously acquired and preprocessed response voltage traveling wave data. This calculation pairs the coded sequence current signal, which serves as a known reference signal, with the response voltage traveling wave data acquired by each inverter node in the collaborative measurement network for analysis. Mathematically, this involves calculating the cross-correlation function of the two discrete-time series:

[0055] ,

[0056] in, It is a measure of the correlation between the encoded sequence current signal and the response voltage traveling wave data collected by the i-th inverter node at a time delay m; It is the value of the coded sequence current signal at time point n, which serves as the reference signal for cross-correlation calculation; This represents the traveling wave response voltage data acquired by the i-th inverter node at time point n+m; m represents the time delay index; and n represents the time series index. This calculation is performed by shifting the reference signal along the time axis. And calculate its relationship with point by point. The sum of their products is used to measure the similarity between the two at different time delays.

[0057] Secondly, peak detection is performed on the cross-correlation coefficient sequence curves of each inverter node. Because the injected coded sequence current signal has a sharp single-peak autocorrelation characteristic, when cross-correlation is performed with the self-echo in the response signal, a very sharp and significant peak will be generated at the corresponding traveling wave propagation delay. The cross-correlation coefficient sequence is automatically searched using an algorithm. The maximum value in the range is determined, and the corresponding time delay index is identified. Multiplying this time delay index by the sampling period yields the initial time delay of the inverter node.

[0058] Finally, a consistency check is performed on the initial time delays calculated for all inverter nodes. This check is based on the distribution network topology and the fundamental physical laws of traveling wave propagation. For example, based on the known physical distance of each inverter node relative to the signal injection point, its initial time delay should exhibit a reasonable gradient relationship, i.e., the farther the node, the longer the delay. The device checks whether the initial time delays of all nodes satisfy this topological constraint. For initial time delay data points that significantly deviate from the physical laws and are inconsistent, the device identifies them as spurious peaks caused by noise or local interference and removes or corrects them. After this check process, the remaining time delay data that is confirmed as valid is finally determined as the propagation time delay of each inverter node for subsequent location calculations.

[0059] S5. Obtain the current line parameters, current meteorological data, and the reference line wave velocity corresponding to the current line, and use the current line parameters and the current meteorological data to correct the reference line wave velocity corresponding to the current line to obtain the real-time wave velocity.

[0060] Optionally, before obtaining the real-time wave velocity, the following steps are also included:

[0061] Acquire historical line parameters, historical meteorological data, historical traveling wave propagation speed, and the corresponding baseline line wave speed for historical lines;

[0062] The difference between the historical traveling wave propagation speed and the reference line wave speed corresponding to the historical line is calculated to obtain the historical wave speed deviation.

[0063] Using the historical route parameters and historical meteorological data as inputs, and the historical wave velocity deviation as output, a neural network model is established and trained to obtain a wave velocity error prediction model.

[0064] Specifically, historical datasets are systematically acquired and organized. These datasets contain four key types of information. First, historical line parameters, covering the physical state of the line at specific points in the past, such as conductor type, erection height, span, and sag. This data is typically obtained from the power grid asset management database. Second, historical meteorological data synchronized with the aforementioned line parameters, including temperature, humidity, air pressure, and rainfall along the line. This data originates from meteorological sensors deployed along the line or authoritative meteorological service agencies. Third, historical traveling wave propagation speed, obtained through post-event analysis of historically occurring fault events with precisely confirmed locations, verified via manual line inspections, or measured through offline calibration experiments. Fourth, the baseline line wave velocity corresponding to each historical moment, a theoretically calculated value primarily determined by the line's inductance and capacitance per unit length, serving as a reference under ideal conditions. Next, this historical data is processed to calculate the historical wave velocity deviation. This deviation is the difference between the actual measured historical traveling wave propagation speed and the theoretically calculated baseline line wave velocity; it quantifies the degree to which the actual wave velocity deviates from the theoretical value under specific line parameters and meteorological conditions. The calculation formula is as follows:

[0065] ,

[0066] in, Represents the historical wave velocity deviation; It is the actual measured value of the historical traveling wave propagation speed obtained by back-calculation from historical failure events or experiments; It is the baseline wave velocity of the historical line calculated based on the theoretical model of the line at that time.

[0067] Finally, a neural network model was built and trained using machine learning techniques. In this model, a multi-dimensional vector composed of historical line parameters and historical meteorological data was used as the model input, while the calculated historical wave velocity deviation was used as the input. The model's expected output is used as the training parameter. By feeding a large number of historical data samples into the neural network and iteratively training it using algorithms such as backpropagation, the model's internal weights and biases are continuously optimized until it can accurately learn the complex nonlinear mapping relationship between the input parameters (route and meteorological conditions) and the output wave speed deviation. After training, the resulting neural network model, which solidifies this mapping law, is the wave speed error prediction model, which can be used for subsequent prediction of real-time wave speed deviation. Figure 3 As shown, the relationship learned by the wave velocity error prediction model is illustrated. For example, the deviation of traveling wave velocity may exhibit a nonlinear relationship with changes in ambient temperature. The scatter points in the figure represent historical measurement data, while the curves represent the relationship predicted by the model.

[0068] Optionally, the method further includes a model pre-training mechanism that combines offline calibration with transfer learning.

[0069] Specifically, for a newly commissioned power line, before acquiring any real fault data, an offline calibration can be performed during planned power outages or low-load windows. Operators carry a portable signal generator and receiver with satellite timing capabilities, interacting with a collaborative measurement network at multiple precisely known geographical locations along the line, such as towers or switches. By measuring the round-trip propagation time of the coded signal between two points at a known distance, the precise wave velocity of that section under current weather and line conditions can be directly calculated. Collecting multiple sets of such distance, time, and environmental data points constitutes a high-quality initial training dataset for the preliminary training of the wave velocity error prediction model, effectively addressing the cold start problem. Secondly, a model initialization strategy based on transfer learning is introduced. For areas with similar line types, voltage levels, and geographical environments, a wave velocity error prediction model already trained on mature lines can be used as the base model. When building a model for a new line, instead of starting training with random weights, this base model is loaded, and then fine-tuned using a limited amount of offline calibration data or a small amount of previously accumulated fault data. Transfer learning can transfer the general laws about the relationship between wave speed and environmental factors learned from the source route to the target new route, thereby reducing the amount of initial training data required, accelerating the convergence speed of the model and improving its generalization ability under small sample conditions.

[0070] Optionally, obtaining the real-time wave velocity includes:

[0071] Obtain current line parameters, current meteorological data, and the corresponding baseline line wave velocity;

[0072] Input the current line parameters and the current meteorological data into the wave speed error prediction model to obtain the current wave speed deviation.

[0073] The current wave velocity deviation is used to correct the wave velocity of the reference line corresponding to the current line, thus obtaining the initial wave velocity.

[0074] The initial wave velocity is calibrated in segments according to the differences in the current line sections to obtain the real-time wave velocity.

[0075] Specifically, the first step is real-time data acquisition. The device obtains current line parameters from the power grid's real-time monitoring system, such as SCADA or line status sensors. These parameters include dynamic physical quantities such as real-time conductor temperature and sag. Simultaneously, it acquires real-time meteorological data about the current line's environment, such as ambient temperature, humidity, and air pressure, through meteorological monitoring stations deployed along the line or by accessing third-party meteorological services. Furthermore, the device calculates a theoretical reference wave velocity based on the current line's static electrical properties, such as inductance and capacitance per unit length. After acquiring the real-time data, the device integrates the current line parameters and meteorological data into an input vector and feeds it into a previously trained wave velocity error prediction model. This model analyzes and calculates the input real-time conditions, outputting a predicted value—the current wave velocity deviation. This deviation represents the expected deviation of the actual propagation speed of the traveling wave from the theoretical reference wave velocity under specific conditions. Then, this predicted deviation is used to correct the reference wave velocity, resulting in a more accurate initial wave velocity.

[0076] ,

[0077] in, Represents the initial wave velocity; It is the reference line wave velocity corresponding to the current line, calculated from the electrical theoretical model of the line; This is the current wave velocity deviation output by the wave velocity error prediction model, corresponding to the current line parameters and meteorological data. Finally, considering that a long distribution network line may consist of different types of cables, overhead lines, or multiple sections traversing different geographical environments, these segment differences will cause the propagation speed of the traveling wave to be not entirely the same in different sections. Therefore, the device performs segmental calibration on the calculated initial wave velocity based on pre-stored line segment information. That is, the initial wave velocity is used as the average correction value for the entire line, and combined with the unique physical properties of each section, such as dielectric constant and geometric structure, the device matches and extracts the corresponding calibration coefficient for each section from a pre-set wave velocity calibration coefficient library. The initial wave velocity is multiplied by the corresponding calibration coefficient for each section to generate the segment wave velocity for each section. The segment wave velocities of all sections and their corresponding section location information are integrated to finally obtain a non-uniform velocity field that accurately reflects the propagation speed of the traveling wave at different locations along the entire line, i.e., the real-time wave velocity.

[0078] S6. Using the propagation time delay and the real-time wave velocity, calculate the fault location data.

[0079] Optionally, the step of calculating the fault location data using the propagation time delay and the real-time wave velocity includes:

[0080] Based on the propagation time delay, calculate the time difference of the traveling wave arriving at each inverter node;

[0081] By combining the time difference and the real-time wave velocity, the preliminary location of the fault point is calculated using a two-end traveling wave positioning method.

[0082] The preliminary location is verified and optimized using redundant data in the collaborative measurement network, thereby outputting fault location data.

[0083] Specifically, the first step is to calculate the time difference between the traveling wave arrival times at different inverter nodes based on the previously extracted propagation time delays of each node. To apply the two-end traveling wave localization method, the device selects a pair of inverter nodes located on either side of the fault point that can clearly receive the traveling wave signal, such as node A and node B. The time difference between the traveling wave arrival times at these two nodes is then calculated. They can be directly affected by their respective propagation time delays. and Subtraction yields, i.e. The propagation time delay here is based on absolute time using a unified spatiotemporal reference, therefore the calculation accuracy of this time difference is very high.

[0084] Subsequently, combining the calculated time difference with the previously corrected real-time wave velocity, the preliminary location of the fault point is calculated using the classical physical model of two-end traveling wave localization. This model calculates the distance from the fault point to a selected node, such as node A. The total length of the line between the two nodes Real-time wave velocity of traveling waves and the time difference between the arrival of the traveling wave at the two nodes. Connect them. The calculation formula is:

[0085] ,

[0086] in, It is the line length from the fault point to node A. It is the total line length between node A and node B, which is a known power grid topology parameter obtained from line design or GIS database; It is the real-time wave velocity obtained previously through the wave velocity error prediction model and segmented calibration, applicable to the line segment between nodes A and B; It is the difference in propagation time delay between the traveling wave arriving at node A and node B. This is calculated using this formula. This is a preliminary location estimate of the fault point.

[0087] Finally, the redundant data in the collaborative measurement network is used to verify and optimize this preliminary location. Since the collaborative measurement network typically contains multiple inverter nodes, the device can select multiple different node pairs, such as A and C, B and C, etc., and repeat the above calculation process to obtain multiple independent estimates of the fault location. Ideally, these estimates should point to the same physical location. By performing consistency comparison and data fusion processing on these multiple preliminary location results, such as using algorithms like weighted average or least squares, outliers can be effectively eliminated, and an optimal estimate can be derived by integrating all valid information. This final result, verified and optimized using multi-source data, is the output fault location data.

[0088] This invention eliminates the errors introduced by the asynchronous clocks of various measuring devices in traditional positioning methods by introducing satellite timing technology and a collaborative measurement network, providing a time reference for the entire positioning process. By employing actively injected coded sequence signals combined with cross-correlation analysis, the detection capability of weak traveling wave signals and the accuracy of wavefront arrival time identification are improved in complex electromagnetic environments, enhancing the device's anti-interference and reliability, making it independent of uncertain transient signals generated by the fault itself. Furthermore, by establishing a real-time wave velocity correction mechanism that considers line parameters and meteorological data, positioning errors caused by wave velocity drift due to environmental changes are overcome, ensuring that the positioning results reflect the real-time status of the line. This method organically integrates high-precision synchronization, active signal detection, and dynamic environment adaptation technologies, achieving high-precision, high-reliability, and robust positioning of distribution network fault locations.

[0089] Optionally, the method further includes:

[0090] Record the propagation time delay, line parameters, and meteorological data of each inverter node when the fault occurs;

[0091] By combining the fault location data and the propagation time delay of each inverter node when the fault occurred, the actual measured value of the line wave velocity during the fault period is calculated.

[0092] The wave velocity error prediction model is corrected using the actual measured values, as well as the line parameters and meteorological data at the time of the fault.

[0093] Specifically, firstly, after the device outputs precise fault location data through the aforementioned steps, it immediately performs a data archiving operation. The device permanently records all relevant data at the moment the fault event occurred, including the propagation time delay measured by each inverter node in the collaborative measurement network, as well as line parameters and meteorological data synchronized with the fault occurrence time acquired through sensors and monitoring systems. Next, the device uses the known fault location results to calculate the actual measured value of the traveling wave propagating on the line during the fault. This calculation utilizes the fundamental physical relationships of distance, velocity, and time. Specifically, for any inverter node i in the network, the actual line distance from it to the determined fault point is calculated. It is known that the total time for the traveling wave to propagate from the injection point to the fault point and then reflect back to node i, or to propagate directly from the injection point to node i, is its corresponding propagation time delay. Therefore, the actual measured value of the average traveling wave propagation speed along this path can be calculated:

[0094] ,

[0095] in, This is the actual measured value of the line wave velocity to be determined; This is the known line distance from inverter node i to the finally determined fault point in this incident. This data is obtained from the power grid GIS system in combination with the fault location results. This is the propagation time delay recorded by inverter node i during this event. By performing this calculation on data from multiple inverter nodes and averaging it, a highly reliable actual measurement of the line wave velocity during this fault can be obtained. Finally, this newly obtained real data sample is used to correct the wave velocity error prediction model. The device uses the line parameters and meteorological data at the time of the fault as a new set of input data, and simultaneously uses the calculated actual measurement of the line wave velocity. Alternatively, the deviation of the wave speed from the theoretical reference wave speed can be used as the corresponding accurate output label. This new set of input-output data is then added to the existing neural network training dataset. Subsequently, the device initiates a retraining or incremental learning process for the wave speed error prediction model. Through this process, the model fine-tunes its internal weight parameters based on the newly added data, thereby enabling it to more accurately understand the mapping between the complex relationship between railway lines, meteorological conditions, and wave speed.

[0096] Optionally, the present invention also includes a real-time topology verification function based on signal response.

[0097] Specifically, the actively injected coded signal itself can be used as a probe to quickly verify the current electrical connections before performing fault location calculations. The main control system internally stores a theoretical topology model based on the GIS system. Based on this model, the theoretical propagation time sequence of the traveling wave that should reach each node in the network when a signal is injected from a certain point can be calculated in advance, forming a topological response fingerprint. After the actual signal injection, the system compares the actual propagation time delay sequence measured by each node with this expected topological response fingerprint. If a node fails to receive the signal at the expected time, or if the reception time deviates significantly from the theoretical value, the system can determine that the electrical path between that node and the injection point has changed, for example, due to a circuit breaker being disconnected. Conversely, if a node that theoretically should not receive a direct wave receives a clear signal, it may indicate a temporary line connection or a faulty switch closure. In this way, the system can identify the differences between the current operating topology and the static database in real time before fault location calculation, and update the line distance parameters used for calculation accordingly, or issue an alarm to the operation and maintenance personnel when the topology is uncertain, thereby avoiding location errors caused by topology errors.

[0098] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides a power distribution network fault location device based on a satellite timing inverter, the device comprising:

[0099] The timing network construction module is used to acquire timing signals from the satellite navigation system, decode and generate a unified spatiotemporal reference, and distribute the unified spatiotemporal reference to multiple inverter nodes in the power distribution network to form a collaborative measurement network.

[0100] The signal injection module is used to inject a coded sequence current signal with sharp single-peak autocorrelation characteristics into the distribution network line from the master control inverter node in the cooperative measurement network, based on the unified spatiotemporal reference.

[0101] The response acquisition module is used to synchronously acquire the corresponding response voltage traveling wave data generated by the coded sequence current signal from each inverter node in the collaborative measurement network based on the unified spatiotemporal reference.

[0102] The correlation analysis module is used to perform cross-correlation analysis between the encoded sequence current signal and the response voltage traveling wave data of each inverter node, and extract the propagation time delay of each inverter node.

[0103] The wave velocity correction module is used to acquire the current line parameters, current meteorological data and the reference line wave velocity corresponding to the current line, and to correct the reference line wave velocity corresponding to the current line using the current line parameters and the current meteorological data to obtain the real-time wave velocity.

[0104] The fault location module is used to calculate the fault location data by using the propagation time delay and the real-time wave velocity.

[0105] Example 1

[0106] This invention was applied to a 15-kilometer-long 10kV distribution network line in the suburbs of a city. The line traverses a complex environment, including farmland, forest areas, and residential areas, and is susceptible to weather changes. Traditional fault location methods have significant errors, and manual line inspection is time-consuming and labor-intensive. Five inverter nodes, numbered A, B, C, D, and E, were deployed at key locations along the line. The device first utilizes the built-in satellite navigation system receiving modules of inverter nodes A, B, C, D, and E to acquire and decode BeiDou satellite timing signals, generating a unified spatiotemporal reference. This reference is then distributed to all nodes via power line carrier communication technology, constructing a collaborative measurement network with nanosecond-level time synchronization accuracy. Node A was designated as the master inverter node. During fault location testing, the device recorded location data under different weather conditions and compared it with traditional traveling wave location methods based on fixed wave velocity.

[0107] A six-month field test was conducted during a certain period, during which three line faults and ten artificially simulated short-circuit faults were recorded. Upon each fault occurrence, the device automatically executed the entire process of signal injection, response acquisition, correlation analysis, wave velocity correction, and location calculation.

[0108] Taking a single-phase ground fault occurring at 9:15 AM on a certain day as an example, the fault occurred between line sections B and C. After the device detected the fault, the main control inverter node A immediately injected a preset m-sequence coded current signal into the line under a unified spatiotemporal reference. All inverter nodes in the collaborative measurement network collected response voltage data containing traveling wave characteristics within a synchronized sampling time window. After preprocessing, the correlation analysis module performed cross-correlation calculations between the injected coded sequence and the response voltage traveling wave data of each node, successfully identifying sharp peaks on the cross-correlation curve. The extracted propagation time delay data of each node was confirmed to be valid after consistency verification.

[0109] To calculate the real-time wave velocity, the device first acquired the prevailing meteorological data: ambient temperature 25°C, humidity 70%, and line parameters. Based on a wave velocity error prediction model trained using historical data, the device predicted the wave velocity deviation under the current conditions as follows: The device modifies this deviation to the reference line wave velocity. Corrections were made, and segmented calibration was performed based on the line characteristics of section BC. The final real-time wave velocity for this section was obtained as follows: .

[0110] Using the propagation time delay data of nodes B and C, which are 55.21 μs and 48.15 μs respectively, and the calculated real-time wave velocity, the initial fault location was calculated to be 4.81 km from node B using the two-end traveling wave localization formula. Simultaneously, the device performed cross-validation using data from the redundant node D, yielding a result of 4.82 km, showing a high degree of consistency between the two calculations. The device ultimately outputs the fault location data as 4.815 km from node B. (Note: The last sentence about using a fixed wave velocity seems unrelated and likely refers to a different calculation method.) The traditional method calculated the location to be 4.92 kilometers, with an error exceeding 100 meters. Repair personnel, using the location provided by this invention, quickly arrived at the scene and discovered that the fault was actually caused by a fallen tree approximately 4.81 kilometers from node B.

[0111] After the fault was resolved, the device used the confirmed fault location data, recorded propagation time delay, line parameters, and meteorological data from this event as a new training sample. The actual wave velocity during the fault was calculated using inverse calculation. Furthermore, this data was used to incrementally learn the wave velocity error prediction model, further improving the model's prediction accuracy.

[0112] Example 2

[0113] This invention is applied to a 10kV distribution network in an industrial park. The network includes a main feeder L1 and a branch line L2 supplying power to customer A. An automatic recloser SW1 is installed on branch line L2. A total of eight satellite timing inverters equipped with the functions of this invention are deployed along the main feeder and branch line, forming a collaborative measurement network. The inverter node M located at the beginning of the main feeder is designated as the master control node. This embodiment aims to verify the positioning accuracy and effectiveness of this invention in a branch network structure, and to demonstrate its real-time topology verification function.

[0114] One afternoon, the park experienced a thunderstorm, and the system detected a fault on branch line L2 belonging to Company A. The recloser SW1 on the line tripped, and after one reclosing, it tripped again and was permanently blocked, indicating that the fault was permanent. The dispatch center then activated the fault location device described in this invention.

[0115] First, the device performed a "real-time topology verification" function. Under a unified spatiotemporal reference, the master inverter node M injected a pre-defined m-sequence encoded current signal into the line. The device compared the actual signal propagation time delay sequence collected by all inverter nodes in the network with the theoretical topology "response fingerprint" pre-stored in the power grid GIS database. The comparison revealed that all inverter nodes located on branch line L2, after recloser SW1, did not receive the probe signal, even though they should theoretically have. The system immediately deduced that recloser SW1 was in an open state, consistent with the switch status information sent by the power grid SCADA system. Based on this, the system confirmed the current real-time operating topology and narrowed the fault range to the branch line L2 section after recloser SW1, avoiding location failure due to incorrect topology information.

[0116] The fault was subsequently identified as a high-impedance grounding fault caused by a broken conductor falling onto a dry tree branch, with a fault resistance as high as 600Ω. The natural traveling wave signal generated at the moment of the fault was extremely weak and completely drowned out by the background noise, causing the traditional passive traveling wave ranging device installed in the substation to fail to trigger effectively. However, this invention, by actively injecting a coded sequence current signal with good autocorrelation characteristics and performing cross-correlation analysis at each synchronization node, still managed to extract the propagation time delay peak from the response signal even under extremely low signal-to-noise ratio conditions, demonstrating excellent detection capability for high-impedance faults.

[0117] Next, the device enters the fault precise location calculation stage. Since the fault is located on a branch line, the device selects the propagation time delay data of three nodes: inverter nodes P1 and P2 on the main feeder located on both sides of the branch point T, and inverter node P3 on the branch line located between the recloser SW1 and the fault point. In the real-time wave velocity correction stage, the device acquires the current meteorological data (temperature 31℃, humidity 90%, thunderstorms) and line operating parameters. Using the trained wave velocity error prediction model, it calculates the real-time wave velocities of the main feeder L1 and the branch line L2 (whose conductor type is different from L1). and .

[0118] Using the precise propagation time delays of these three nodes and the corrected segmented real-time wave velocities, the device established a hyperbolic positioning equation system based on multi-point measurements. By solving this equation system, a unique location solution satisfying all time constraints was obtained. The calculation results show that the fault point is located on branch line L2, 5.12 kilometers away from the branch point T (i.e., the connection point between L1 and L2).

[0119] Based on the location result, maintenance personnel went directly to the target location and found the broken conductor hanging on a tree branch near a tower approximately 5.1 kilometers from the branch point. The actual location was only 15 meters off from the initial location. This invention not only achieves high-precision positioning in simple linear networks, but also, in more challenging branch network topologies, can quickly and accurately pinpoint fault locations through real-time topology verification and multi-point collaborative measurement. Furthermore, its active detection mode effectively overcomes the limitations of traditional methods in handling complex faults such as high-impedance grounding, improving the efficiency and reliability of distribution network fault handling.

[0120] In summary, the method of this invention can dynamically correct the traveling wave velocity based on real-time meteorological conditions, providing a foundation for accurate positioning. The time synchronization and dynamic wave velocity correction of this invention shorten fault diagnosis time and improve power supply reliability. This invention possesses continuous self-optimization capabilities, adapting to line aging and seasonal environmental changes, maintaining positioning performance over the long term.

[0121] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any method of indirect connection is applicable to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0122] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A method for fault location in distribution networks based on satellite time-synchronized inverters, characterized in that, The method includes: The timing signal from the satellite navigation system is acquired, decoded to generate a unified spatiotemporal reference, and the unified spatiotemporal reference is distributed to multiple inverter nodes in the power distribution network to form a collaborative measurement network. Based on the unified spatiotemporal reference, the master control inverter node in the collaborative measurement network injects a coded sequence current signal with sharp single-peak autocorrelation characteristics into the distribution network line; Based on the unified spatiotemporal reference, each inverter node in the collaborative measurement network synchronously collects the corresponding response voltage traveling wave data generated by the encoded sequence current signal; The propagation time delay of each inverter node is extracted by performing cross-correlation analysis between the encoded sequence current signal and the traveling wave data of the response voltage of each inverter node. This includes: performing cross-correlation calculation between the encoded sequence current signal and the traveling wave data of the response voltage of each inverter node to generate cross-correlation coefficient sequence curves for each inverter node; identifying the peak positions in the cross-correlation coefficient sequence curves of each inverter node to extract the initial time delay of each inverter node; performing consistency verification on the initial time delay of each inverter node, and generating the propagation time delay of each inverter node based on the verification results. The process involves acquiring current line parameters, current meteorological data, and the corresponding baseline line wave velocity, and then using the current line parameters and current meteorological data to correct the baseline line wave velocity to obtain the real-time wave velocity. This includes: acquiring the current line parameters, current meteorological data, and the corresponding baseline line wave velocity; inputting the current line parameters and current meteorological data into a wave velocity error prediction model to obtain the current wave velocity deviation; using the current wave velocity deviation to correct the corresponding baseline line wave velocity to obtain the initial wave velocity; performing segmented calibration of the initial wave velocity based on the segment differences of the current line to obtain the real-time wave velocity; and using the propagation time delay and the real-time wave velocity to calculate the fault location data. The construction of the wave velocity error prediction model includes: acquiring historical line parameters, historical meteorological data, historical traveling wave propagation speed, and the reference line wave velocity corresponding to the historical line; calculating the difference between the historical traveling wave propagation speed and the reference line wave velocity corresponding to the historical line to obtain the historical wave velocity deviation; and establishing and training a neural network model with the historical line parameters and historical meteorological data as inputs and the historical wave velocity deviation as output to obtain the wave velocity error prediction model.

2. The method for fault location in a distribution network based on a satellite time-synchronized inverter according to claim 1, characterized in that, The collaborative measurement network comprises: The timing signal from the satellite navigation system is acquired and decoded to obtain time synchronization information and position information. By integrating the time synchronization information and the location information, a unified spatiotemporal reference is generated; The unified spatiotemporal reference is distributed to each inverter node via power line carrier communication or fiber optic network to form a collaborative measurement network.

3. The method for fault location in a distribution network based on a satellite time-synchronized inverter according to claim 2, characterized in that, The process of injecting a coded sequence current signal with sharp single-peak autocorrelation characteristics into the distribution network lines by the master control inverter node in the collaborative measurement network, based on the unified spatiotemporal reference, includes: Based on the unified spatiotemporal reference, a fundamental wave signal of a coded sequence with sharp single-peak autocorrelation characteristics is generated; The encoded sequence current signal is obtained by modulating the fundamental wave signal of the encoded sequence and then injected into the distribution network line through the master control inverter node in the collaborative measurement network.

4. The method for fault location in a distribution network based on a satellite time-synchronized inverter according to claim 3, characterized in that, The step of synchronously acquiring the corresponding response voltage traveling wave data generated by the encoded sequence current signal based on the unified spatiotemporal reference by each inverter node in the collaborative measurement network includes: Based on the unified spatiotemporal reference, a sampling time window is configured for each inverter node; Within the sampling time window, raw response voltage traveling wave data generated by the coded sequence current signal are acquired; The original response voltage traveling wave data is preprocessed to remove power frequency components and high-frequency noise, thereby generating response voltage traveling wave data.

5. The method for fault location in a distribution network based on a satellite time-synchronized inverter according to claim 4, characterized in that, The calculation of the fault location data using the propagation time delay and the real-time wave velocity includes: Based on the propagation time delay, calculate the time difference of the traveling wave arriving at each inverter node; By combining the time difference and the real-time wave velocity, the preliminary location of the fault point is calculated using a two-end traveling wave positioning method. The preliminary location is verified and optimized using redundant data in the collaborative measurement network, thereby outputting fault location data.

6. The method for fault location in a distribution network based on a satellite time-synchronized inverter according to claim 5, characterized in that, The method further includes: Record the propagation time delay, line parameters, and meteorological data of each inverter node when the fault occurs; By combining the fault location data and the propagation time delay of each inverter node when the fault occurred, the actual measured value of the line wave velocity during the fault period is calculated. The wave velocity error prediction model is corrected using the actual measured values, as well as the line parameters and meteorological data at the time of the fault.

7. A power distribution network fault location device based on a satellite time-synchronized inverter, characterized in that, The device includes: The timing network construction module is used to acquire timing signals from the satellite navigation system, decode and generate a unified spatiotemporal reference, and distribute the unified spatiotemporal reference to multiple inverter nodes in the power distribution network to form a collaborative measurement network. The signal injection module is used to inject a coded sequence current signal with sharp single-peak autocorrelation characteristics into the distribution network line from the master control inverter node in the cooperative measurement network, based on the unified spatiotemporal reference. The response acquisition module is used to synchronously acquire the corresponding response voltage traveling wave data generated by the coded sequence current signal from each inverter node in the collaborative measurement network based on the unified spatiotemporal reference. The correlation analysis module is used to perform cross-correlation analysis between the encoded sequence current signal and the response voltage traveling wave data of each inverter node to extract the propagation time delay of each inverter node. This includes: performing cross-correlation calculation between the encoded sequence current signal and the response voltage traveling wave data of each inverter node to generate a cross-correlation coefficient sequence curve for each inverter node; identifying the peak position in the cross-correlation coefficient sequence curve of each inverter node to extract the initial time delay of each inverter node; performing consistency verification on the initial time delay of each inverter node; and generating the propagation time delay of each inverter node based on the verification results. The wave velocity correction module is used to acquire current line parameters, current meteorological data, and the wave velocity of the reference line corresponding to the current line, and to correct the wave velocity of the reference line corresponding to the current line using the current line parameters and the current meteorological data to obtain the real-time wave velocity. This includes: acquiring the current line parameters, current meteorological data, and the wave velocity of the reference line corresponding to the current line; inputting the current line parameters and the current meteorological data into a wave velocity error prediction model to obtain the current wave velocity deviation; correcting the reference line wave velocity corresponding to the current line using the current wave velocity deviation to obtain the initial wave velocity; and performing segmented calibration of the initial wave velocity according to the segment differences of the current line to obtain the real-time wave velocity. The fault location module is used to calculate the fault location data using the propagation time delay and the real-time wave velocity; The construction of the wave velocity error prediction model includes: acquiring historical line parameters, historical meteorological data, historical traveling wave propagation speed, and the reference line wave velocity corresponding to the historical line; calculating the difference between the historical traveling wave propagation speed and the reference line wave velocity corresponding to the historical line to obtain the historical wave velocity deviation; and establishing and training a neural network model with the historical line parameters and historical meteorological data as inputs and the historical wave velocity deviation as output to obtain the wave velocity error prediction model.

Images