Flexible capacity expansion system fault positioning method and system based on traveling wave injection

Abstract

The invention relates to the technical field of power grid fault positioning, and discloses a flexible capacity expansion system fault positioning method and system based on traveling wave injection, and the method comprises the steps: determining a test traveling wave based on the attribute of a node in a flexible capacity expansion system; determining an injection point of a test traveling wave based on the attribute of the node; transmitting a test traveling wave from the injection point, and obtaining a real-time feedback echo based on a preset monitoring point; determining a first fault mode of the flexible capacity expansion system through a pre-trained graph neural network model according to the test traveling waves and the attributes of the nodes; determining a second fault mode of the flexible capacity expansion system according to the real-time feedback echo; and determining a fault node of the flexible capacity expansion system based on the first fault mode and the second fault mode. The method can adapt to the complex topology and dynamic operation characteristics of the flexible capacity expansion system, improves the accuracy of fault node positioning, and improves the maintenance efficiency of a power grid.

Description

Technical Field

[0001] This application relates to the field of power grid fault location technology, specifically to a fault location method and system for a flexible capacity expansion system based on traveling wave injection. Background Technology

[0002] As the energy structure shifts towards low-carbon and distributed energy, flexible expansion systems containing distributed power sources and energy storage systems have become an important vehicle for improving power supply reliability. However, the system has a complex topology and diverse node attributes, making rapid and accurate node location after a fault a critical requirement.

[0003] Currently, existing fault location technologies for flexible capacity expansion systems mainly rely on traditional impedance methods, natural traveling wave methods, and conventional traveling wave injection methods. However, these methods have several shortcomings. For example, traditional impedance methods are greatly affected by operating conditions and transition resistance, resulting in low location efficiency. Natural traveling wave methods rely on the natural traveling wave of the fault, which has weak signals and poor noise immunity. While conventional traveling wave injection methods actively transmit traveling waves, they do not incorporate node attributes into the design of testing the traveling wave and selecting the injection point. Furthermore, they rely on a single method to determine the fault mode, failing to integrate the correlation between the traveling wave and node attributes, as well as real-time echo information, making it difficult to accurately determine the fault node and thus unable to meet the system's location requirements. Therefore, existing technologies suffer from insufficient accuracy in fault node location and low grid maintenance efficiency.

[0004] The preceding description is intended to provide general background information and does not necessarily constitute prior art. Summary of the Invention

[0005] This application provides a fault location method and system for flexible capacity expansion systems based on traveling wave injection. It can adapt to the complex topology and dynamic operating characteristics of flexible capacity expansion systems, improve the accuracy of fault node location, and thus improve power grid maintenance efficiency.

[0006] Based on this, embodiments of this application provide a fault location method for a flexible capacity expansion system based on traveling wave injection, including: Based on the attributes of nodes in the flexible expansion system, the test traveling wave is determined; Based on the attributes of the node, the injection point of the test traveling wave is determined; The test traveling wave is emitted from the injection point, and real-time feedback echoes are obtained based on pre-set monitoring points; Based on the properties of the test traveling wave and the node, the first failure mode of the flexible expansion system is determined using a pre-trained graph neural network model. Based on the real-time feedback echo, the second fault mode of the flexible capacity expansion system is determined; Based on the first fault mode and the second fault mode, the fault node of the flexible expansion system is determined.

[0007] Furthermore, in some embodiments of this application, determining the test traveling wave based on the attributes of nodes in the flexible capacity expansion system includes: Obtain the attributes of the node, including at least one of electrical parameters, connection relationships, and operating status; Based on the attributes of the node, adjust the frequency and amplitude of the test traveling wave; Based on the topological location of the node and its neighboring nodes, design the waveform and encoding method of the test traveling wave.

[0008] Furthermore, in some embodiments of this application, adjusting the frequency and amplitude of the test traveling wave based on the attributes of the node includes: Based on the voltage level and load type of the node, the initial traveling wave frequency and initial amplitude parameters are generated; Based on the initial traveling wave frequency and the initial amplitude parameters, and in conjunction with the real-time operating status of the node, the optimized traveling wave emission parameters are calculated. Based on the optimized traveling wave transmission parameters, the test traveling wave is configured and transmitted.

[0009] Furthermore, in some embodiments of this application, determining the injection point of the test traveling wave based on the attributes of the node includes: Based on the electrical parameters of the nodes, nodes with high voltage levels or short-circuit capacity greater than a preset capacity are selected as candidate injection points. Based on the topological location of the node, a hub node connecting multiple lines is preferentially selected as the injection point; The selection of injection points is dynamically adjusted based on the operating status and fault history of the nodes.

[0010] Furthermore, in some embodiments of this application, the step of emitting the test traveling wave from the injection point and obtaining real-time feedback echoes based on pre-set monitoring points includes: Based on the electrical characteristics and topological location of the nodes, corresponding monitoring points are set; A traveling wave signal acquisition device is deployed at the monitoring point to capture the reflection and refraction signals of the test traveling wave in real time; The collected reflected and refracted signals are preprocessed to generate real-time feedback echoes.

[0011] Furthermore, in some embodiments of this application, determining the first failure mode of the flexible capacity expansion system based on the test traveling wave and the attributes of the node using a pre-trained graph neural network model includes: The test traveling wave is fused with the attributes of the node to generate a node feature vector containing the traveling wave characteristics; Based on the node feature vectors and the topological connection relationships of the flexible expansion system, a graph structure data is constructed. The graph structure data is input into the pre-trained graph neural network model to obtain the hidden feature representation of the node; Based on the hidden feature representation of the node, the first fault mode is output, which is the fault probability distribution of each node in the flexible expansion system.

[0012] Furthermore, in some embodiments of this application, fusing the test traveling wave with the attributes of the node to generate a node feature vector containing traveling wave features includes: The waveform features of the test traveling wave at multiple monitoring points are extracted to obtain a traveling wave feature set, wherein the waveform features include arrival time, amplitude and polarity; The traveling wave feature set is associated with and concatenated with the attributes of the node to construct the original node feature vector; The original node feature vector is denoised and normalized to generate a node feature vector containing traveling wave characteristics.

[0013] Furthermore, in some embodiments of this application, determining the second fault mode of the flexible capacity expansion system based on the real-time feedback echo includes: The real-time feedback echoes are preprocessed and feature extracted to obtain the echo feature vectors of each monitoring point; The echo feature vectors from each monitoring point are integrated, and combined with the topology of the flexible expansion system, to generate a system-level echo feature matrix. Based on the system-level echo feature matrix, the second fault mode is determined by analyzing the time difference of arrival and propagation path of the traveling wave. The second fault mode includes the location and type of the fault point.

[0014] Furthermore, in some embodiments of this application, determining the fault node of the flexible capacity expansion system based on the first fault mode and the second fault mode includes: Based on the first fault mode and the second fault mode, a fusion decision factor is generated, which is used to quantify the degree of consistency between the first fault mode and the second fault mode in terms of fault location and attributes. Based on the fusion decision factors, the final failure node of the flexible expansion system is determined; When the fusion decision factor is lower than a preset threshold, an optimization feedback signal is triggered for the test wave injection strategy or the graph neural network model.

[0015] Accordingly, this application also provides a fault location system for a flexible capacity expansion system based on traveling wave injection, comprising: The test traveling wave module is used to determine the test traveling wave based on the attributes of nodes in a flexible expansion system; The injection point module is used to determine the injection point of the test traveling wave based on the attributes of the node. The feedback echo module is used to emit the test traveling wave from the injection point and acquire real-time feedback echoes based on pre-set monitoring points; The first fault module is used to determine the first fault mode of the flexible expansion system based on the test traveling wave and the attributes of the node through a pre-trained graph neural network model. The second fault module is used to determine the second fault mode of the flexible expansion system based on the real-time feedback echo. The fault determination module is used to determine the fault node of the flexible expansion system based on the first fault mode and the second fault mode.

[0016] This application provides a fault location method and system for a flexible capacity expansion system based on traveling wave injection, comprising: determining a test traveling wave based on the attributes of nodes in the flexible capacity expansion system; determining the injection point of the test traveling wave based on the attributes of the nodes; emitting the test traveling wave from the injection point and acquiring real-time feedback echoes based on pre-set monitoring points; determining a first fault mode of the flexible capacity expansion system based on the test traveling wave and the attributes of the nodes through a pre-trained graph neural network model; determining a second fault mode of the flexible capacity expansion system based on the real-time feedback echoes; and determining the faulty node of the flexible capacity expansion system based on the first fault mode and the second fault mode. The fault location scheme for flexible capacity expansion systems based on traveling wave injection provided in this application firstly generates a detection signal with controllable intensity and characteristics by dynamically determining and injecting test traveling waves based on node attributes, thus overcoming the shortcomings of relying on weak natural fault traveling waves from the source. Next, graph neural networks are used in parallel to analyze the deep correlation between the power grid topology and traveling wave propagation, and the actual physical response of the fault is captured through monitoring point echoes, thereby forming a first fault mode based on model inference and a second fault mode based on real-time signal analysis. Finally, these two fault modes, derived from different principles and complementary in nature, are fused for decision-making and cross-validation to determine the fault node of the flexible capacity expansion system. Therefore, this application can improve the accuracy of fault node location in flexible capacity expansion systems, solve the location error problem that may be caused by relying on a single signal source or a single judgment method, achieve precise location of fault nodes in flexible capacity expansion systems, and thus improve power grid maintenance efficiency. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is an application environment diagram of the fault location method for flexible capacity expansion systems based on traveling wave injection provided in the embodiments of this application; Figure 2 This is a flowchart illustrating the fault location method for a flexible capacity expansion system based on traveling wave injection provided in an embodiment of this application. Figure 3 This is a schematic diagram of the fault location system of the flexible capacity expansion system based on traveling wave injection provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0019] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of systems and methods consistent with those detailed in the appended claims or with some aspects of this application.

[0020] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover descriptions such as non-exclusive inclusion, so that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, components, features, and elements with the same names in different embodiments of this application may have the same meaning or different meanings, the specific meaning of which must be determined by its interpretation in that specific embodiment or further in conjunction with the context of that specific embodiment.

[0021] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0022] In the following description, the use of suffixes such as "module," "part," or "unit" to denote elements is solely for the purpose of illustrative purposes and has no specific meaning in itself. Therefore, "module," "part," or "unit" may be used interchangeably.

[0023] To address the aforementioned technical problems and overcome the shortcomings of existing technologies, this application provides a fault location method and system for flexible capacity expansion systems based on traveling wave injection. This method can adapt to the complex topology and dynamic operating characteristics of flexible capacity expansion systems, improve the accuracy of fault node location, and thus improve power grid maintenance efficiency.

[0024] Figure 1 This is an application environment diagram of a fault location method for a flexible capacity expansion system based on traveling wave injection in one embodiment. (Refer to...) Figure 1 This wave-injection-based fault location method for flexible capacity expansion systems is applied to a wave-injection-based fault location system. The wave-injection-based flexible capacity expansion system fault location system includes a terminal 110 and a server 120. The terminal 110 and server 120 are connected via a network. The terminal 110 can be a desktop terminal or a mobile terminal, and the mobile terminal can be at least one of a mobile phone, tablet computer, or laptop computer. The server 120 can be a standalone server or a server cluster composed of multiple servers. The server 120 is configured to execute the above-mentioned wave-injection-based fault location method for flexible capacity expansion systems, including: determining a test wave based on the attributes of nodes in the flexible capacity expansion system; determining the injection point of the test wave based on the attributes of the nodes; emitting the test wave from the injection point and obtaining real-time feedback echoes based on pre-set monitoring points; determining a first fault mode of the flexible capacity expansion system based on the test wave and node attributes using a pre-trained graph neural network model; determining a second fault mode of the flexible capacity expansion system based on the real-time feedback echoes; and determining the faulty node of the flexible capacity expansion system based on the first and second fault modes.

[0025] Please see Figure 2 , Figure 2 This is a flowchart illustrating a fault location method for a flexible capacity expansion system based on traveling wave injection, provided in an embodiment of this application. This embodiment primarily uses the application of this method to computer equipment as an example. Specifically, the fault location method for a flexible capacity expansion system based on traveling wave injection provided in an embodiment of this application may include the following steps: S1. Determine the test traveling wave based on the attributes of the nodes in the flexible expansion system; Specifically, for step S1, the nodes in the flexible capacity expansion system have specific attributes. These attributes are the core basis for determining the test traveling wave. These node attributes include, but are not limited to, the node's electrical parameters (such as voltage level and short-circuit capacity), connection relationships (such as topological location, number and type of adjacent nodes), operating status (such as load weight and online status), and the load type corresponding to the node (such as residential load, commercial load, and industrial load). The essence of determining the test traveling wave is to ensure that the traveling wave characteristics match the node attributes, guaranteeing that the traveling wave can effectively propagate in the system and accurately reflect fault information. This avoids problems such as excessively rapid signal attenuation, interference with normal system operation, or inability to capture fault characteristics due to mismatch between the traveling wave and node attributes.

[0026] S2. Determine the injection point of the test traveling wave based on the node's attributes; Specifically, for step S2, the selection of the injection point needs to rely on the node attributes. The core is to ensure that the injection point can maximize the coverage of the flexible expansion system by the test traveling wave, and that the signal stability is high during the traveling wave propagation process. This avoids problems such as incomplete traveling wave coverage, signal susceptibility to interference, or inability to reach critical areas due to improper injection point selection. When selecting, it is necessary to comprehensively judge the node's electrical parameters (e.g., high-voltage level nodes have low traveling wave propagation loss, and nodes with large short-circuit capacity can better maintain the traveling wave signal), connection relationships (e.g., hub nodes connect multiple lines, and the traveling wave can propagate in multiple directions), and operating status (e.g., injecting a traveling wave into a node with a light load has little impact on the normal operation of the system).

[0027] S3. Emit a test traveling wave from the injection point and obtain real-time feedback echoes based on pre-set monitoring points; Specifically, for step S3, firstly, according to the determined test traveling wave characteristics (such as waveform, frequency, and amplitude), a test traveling wave is actively transmitted from the injection point determined in step 2 to the flexible expansion system. Secondly, monitoring points are pre-set in the system. These monitoring points are distributed at key locations (such as along the line and near important nodes) that can capture the propagation, reflection, and refraction signals of the traveling wave. Their function is to collect in real time the reflected and refracted waves generated when the traveling wave encounters a fault point or changes in line characteristics during propagation. These collected reflected and refracted waves are the real-time feedback echoes. This step, by actively transmitting the traveling wave and collecting the echoes, provides direct on-site signal evidence for subsequent fault diagnosis.

[0028] S4. Based on the properties of the test wave and nodes, determine the first failure mode of the flexible expansion system using a pre-trained graph neural network model; Specifically, for step S4, firstly, the key information of the determined test traveling wave (such as the propagation characteristics and amplitude variation trend of the traveling wave) is associated and integrated with the node attributes (such as node voltage level, load type, and topological location) to form input data that the model can recognize; secondly, the input data is input into a pre-trained graph neural network model, which has learned the correlation between "test traveling wave-node attributes" and "fault state" through training, and can make a preliminary judgment on the system fault situation; the final output first fault mode is essentially a theoretical judgment result on the system fault formed by the model's analysis of the system's inherent attributes and active detection signals, which is usually presented in the form of the fault probability distribution of each node (i.e., the probability of each node failing).

[0029] S5. Determine the second fault mode of the flexible capacity expansion system based on real-time feedback echoes; Specifically, for step S5, the real-time feedback echo contains key information reflecting fault characteristics, such as the echo arrival time, amplitude, polarity (positive / negative), and rise time. This information is the core basis for determining the second fault mode. By analyzing the echo arrival time difference (the time difference in echo capture by different monitoring points) and combining it with the propagation speed of the traveling wave in the system, the approximate location of the fault point can be preliminarily determined. By analyzing the echo amplitude, polarity, and rise time, the fault type can be further determined. For example, short-circuit faults usually correspond to large echo amplitudes and short rise times, while ground faults usually correspond to echoes of a specific polarity. Finally, the second fault mode is output. The second fault mode is a practical judgment result of the system fault based on the echo signals collected in real time on site, and usually includes the specific location of the fault point and the fault type.

[0030] S6. Based on the first and second failure modes, determine the fault nodes of the flexible capacity expansion system; Specifically, for step S6, the first fault mode is a theoretical fault judgment based on the model and system attributes, and the second fault mode is a practical fault judgment based on real-time echoes. The fusion of the two can compensate for the deviation of the single mode judgment. If the first fault mode points to a node with a high probability of failure, and the fault location and fault type determined by the second fault mode match the attributes of the node (such as the line it is located on and the load characteristics), then the node can be identified as a fault node. If there is a slight difference between the two, the fault node can be further locked by verifying the degree of consistency between the two in terms of fault location and fault type, thus avoiding misjudgment or omission caused by relying solely on theory or real-time data.

[0031] This embodiment first determines the test traveling wave and injection point based on node attributes, ensuring that the traveling wave characteristics are compatible with the system characteristics from the source, avoiding problems such as poor traveling wave propagation and incomplete coverage caused by untargeted design. Second, it actively transmits traveling waves and collects real-time echoes, overcoming the limitations of traditional reliance on weak and noise-resistant natural traveling wave signals, and obtaining stable fault-related signals. Third, the first fault mode relies on a pre-trained graph neural network to achieve theoretical fault judgment by combining the test traveling wave and node attributes, while the second fault mode relies on real-time echoes to achieve practical fault judgment, thus forming a first fault mode based on model inference and a second fault mode based on real-time signal analysis. Finally, the two fault modes, which are derived from different principles and are complementary, are fused for decision-making. Essentially, this involves cross-validating the structured prior knowledge of the power grid with the instantaneous physical evidence of the fault occurrence. This process can effectively correct misjudgments by single methods, thereby overcoming noise interference and path ambiguity as a whole, and achieving stable and accurate fault location.

[0032] Furthermore, in some embodiments, step S1, "determining the test traveling wave based on the attributes of nodes in the flexible expansion system," may specifically include: S11. Obtain the node's attributes, which include at least one of electrical parameters, connection relationships, and operating status; S12. Adjust the frequency and amplitude of the test traveling wave based on the node's attributes; S13. Based on the topological location of the node and its neighboring nodes, design the waveform and encoding method for testing the traveling wave.

[0033] Furthermore, in some embodiments, step S12, "adjusting the frequency and amplitude of the test traveling wave based on the node's attributes," may specifically include: S121. Generate initial traveling wave frequency and initial amplitude parameters based on the node's voltage level and load type; S122. Based on the initial traveling wave frequency and initial amplitude parameters, and combined with the real-time operating status of the node, the optimized traveling wave transmission parameters are calculated. S123. Configure and transmit a test traveling wave based on optimized traveling wave transmission parameters.

[0034] Specifically, step S1 primarily involves determining the test traveling wave of the flexible capacity expansion system based on the attributes of the nodes within the system. A node in the flexible capacity expansion system is a basic unit in the power system, responsible for converting high-voltage electrical energy into low-voltage electrical energy and distributing it to electricity users within its area. The multiple regions covered by the distribution network correspond to different power supply substations. Nodes in the flexible capacity expansion system correspond to different node types. For example, based on load type, nodes can be categorized into four types: urban residential nodes, commercial nodes, industrial nodes, and agricultural nodes. These different types of nodes differ in terms of electricity consumption characteristics, load distribution, and power supply demand. Furthermore, different regions correspond to different types of renewable energy sources, thus different regions correspond to different renewable energy attributes.

[0035] In specific embodiments, nodes can be devices such as transformers or user loads, representing independent entities in the flexible capacity expansion system. Nodes are connected via transmission lines to form the topology of the transmission network. Specifically, abstracting the power nodes in the flexible capacity expansion system into graph-structured nodes includes: determining the node's attribute data, and determining node attributes based on the attribute data. The node's attribute data includes physical attributes, electrical attributes, and operating status. Specifically, physical attributes include the power node's location, capacity, and load; electrical attributes include voltage level, current, and power; operating status includes whether it is online or in a fault state; attribute data also includes node number, load level, and renewable energy access capacity. Attribute data also includes node type: residential, commercial, industrial, and agricultural. Numerical attributes include output power, output current, output voltage, supplied area, and supplied population.

[0036] In a specific embodiment, a transmission line is the physical medium connecting different distribution substations and is responsible for transmitting electrical energy. Therefore, it can be abstracted as an edge to represent the connection relationship between nodes. Edge attributes include line length, impedance, and maximum allowable current. Transmission lines or distribution lines are defined as edges. Since current flows bidirectionally in transmission lines, the edge between any two power supply substations only needs to represent the connection relationship and can be defined as an undirected edge.

[0037] The attributes of an edge include at least physical, electrical, and state attributes. Physical attributes include line length, material, and cross-sectional area; electrical attributes include resistance, reactance, and current carrying capacity (ampere capacity); and state attributes include whether it is online, its fault status, and its degree of aging. In flexible capacity expansion systems, node attributes contain rich information and are crucial for designing injected traveling waves. Injected traveling waves can be designed based on the node's electrical parameters, connection relationships, and operating status.

[0038] In specific embodiments, the injected traveling wave can be designed based on the node's electrical parameters. Specifically, the node voltage level can be considered, as different voltage levels significantly affect the propagation and reflection characteristics of the traveling wave. For high-voltage nodes, the attenuation of the traveling wave during propagation is relatively small, but it may be subject to more complex electromagnetic interference. Therefore, when designing the injected traveling wave, the energy of the traveling wave should be increased to ensure that it can effectively propagate and be accurately detected in long-distance, high-voltage lines. Simultaneously, to avoid causing additional impact on the high-voltage system, the frequency and amplitude of the traveling wave need to be reasonably controlled to match the electrical characteristics of the system. For example, for nodes of ultra-high voltage transmission lines, low-frequency, high-amplitude traveling waves can be used, which can reduce the attenuation of the traveling wave during propagation and reduce the impact on system stability. For low-voltage nodes, due to the shorter line and lower impedance, the propagation speed and reflection characteristics of the traveling wave differ from those of high-voltage lines. In this case, high-frequency, low-amplitude traveling waves can be designed to improve the sensitivity of fault detection while avoiding damage to low-voltage equipment due to excessive traveling wave energy.

[0039] In specific embodiments, the traveling wave can be designed according to the type and size of the node load. The type (e.g., resistive, inductive, capacitive loads) and size of the loads connected to the node will change the equivalent impedance of the node, thus affecting the propagation and reflection of the traveling wave. For nodes mainly connected to inductive loads, the frequency of the injected traveling wave should avoid the resonant frequency of the inductive load to prevent resonance amplification, which could damage the equipment or interfere with the detection of the traveling wave signal. For example, if a node is connected to many large motors, which will exhibit inductive characteristics during startup or operation, a mid-to-high frequency traveling wave can be injected to reduce interaction with the inductive load. Simultaneously, the energy of the traveling wave should be adjusted according to the node load size. Nodes with larger loads require stronger traveling wave signals to effectively penetrate the load's influence and accurately detect fault reflection waves. For nodes with smaller loads, the traveling wave energy can be appropriately reduced to avoid energy waste and unnecessary interference.

[0040] In specific implementations, the injected traveling wave can be designed in conjunction with the node connection relationships. Specifically, the topological location of the node is analyzed; the node's position in the power grid topology determines the propagation path and possible reflection points of the traveling wave. For nodes located at power grid hubs and connecting multiple lines, the injected traveling wave will propagate in multiple directions. To comprehensively acquire fault information, a traveling wave with broadband characteristics and good directionality should be designed. Broadband characteristics enable the traveling wave to cover fault characteristics at different frequencies, improving the accuracy of fault detection; good directionality helps distinguish traveling waves from different directions, facilitating accurate fault location determination. For example, when injecting a traveling wave at a substation bus node, a multi-band composite pulse traveling wave can be used, with pulses of different frequencies propagating to different lines. By analyzing the traveling wave signals returned from each line, the line where the fault is located can be determined. For terminal nodes, since the traveling wave propagation path is relatively simple, the focus should be on the energy concentration and anti-interference capability of the traveling wave to ensure that the traveling wave can effectively propagate on the terminal line and accurately detect the fault.

[0041] In specific embodiments, the influence of adjacent nodes can be considered. The electrical characteristics and connection methods of adjacent nodes will affect the propagation of the traveling wave at this node. If power electronic devices or other nonlinear components exist at adjacent nodes, harmonic interference will be generated. In this case, anti-interference coding techniques, such as pseudo-random coding, should be used in the design of the injected traveling wave. By encoding the traveling wave, the receiving end can extract the original injected traveling wave signal from the complex interference signal through decoding algorithms, thereby improving the reliability of the traveling wave signal. At the same time, the frequency and amplitude of the traveling wave can be adjusted according to the connection line length and impedance of adjacent nodes to optimize the propagation effect of the traveling wave between nodes. For example, when adjacent nodes are connected by long-distance, high-impedance lines, the amplitude of the traveling wave can be appropriately increased and the frequency of the traveling wave can be decreased to reduce the attenuation and distortion of the traveling wave during propagation.

[0042] In specific embodiments, the injected traveling wave can be designed based on the node's operating status. Specifically, node operating parameters can be monitored in real time, such as voltage, current, and power, to understand the node's real-time operating status. When a node is operating under heavy load, the system's stability is relatively poor, and the injection of the traveling wave should be more cautious. The amplitude of the traveling wave can be appropriately reduced to avoid excessive voltage fluctuations or frequency shifts caused by the injection, which could affect the normal operation of the system. At the same time, the frequency of the traveling wave should be adjusted to adapt to the electrical characteristics of the system under heavy load, ensuring that the traveling wave can effectively propagate and detect faults. For example, when the node load reaches more than 80% of its rated capacity, the amplitude of the traveling wave can be reduced to 70%-80% of the normal value, and the traveling wave can be injected in a frequency band near the system's natural frequency to improve the accuracy of traveling wave detection.

[0043] In specific implementations, the operating state of nodes may change over time, such as due to the connection and disconnection of distributed power sources or load fluctuations. To ensure accurate fault location, the injected traveling wave should have adaptive adjustment capabilities. The parameters of the traveling wave, such as frequency, amplitude, and encoding method, can be automatically adjusted by monitoring node state changes in real time. For example, when a new distributed power source connects to a node, the system's impedance characteristics will change. In this case, the frequency and amplitude of the traveling wave can be recalculated and adjusted based on the new system parameters to ensure that the traveling wave accurately reflects fault information. Simultaneously, the anti-interference encoding method is updated to cope with new interference that may be introduced by distributed power sources.

[0044] Furthermore, in some embodiments, step S2, "determining the injection point of the test traveling wave based on the node's attributes," may specifically include: S21. Based on the electrical parameters of the nodes, select nodes with high voltage levels or short-circuit capacity greater than the preset capacity as candidate injection points; S22. Based on the topological location of the node, prioritize selecting the hub node that connects multiple lines as the injection point; S23. Dynamically adjust the selection of injection points based on the node's operating status and fault history.

[0045] Specifically, for step S2, the injection point of the test traveling wave is determined based on the attributes of any node in the flexible capacity expansion system. To determine the importance of each node corresponding to a power supply area in the transmission network, when expanding power supply areas in the transmission network, the higher the importance of any node, the higher the priority of the expansion. In the flexible capacity expansion system, node attributes contain rich information, which is crucial for determining the injection point of the test traveling wave. The method for determining the injection point will be analyzed by starting with the node's electrical parameters, connection relationships, operating status, and other attributes, combined with the power grid structure and fault location requirements.

[0046] In specific embodiments, the injection point can be determined based on the node's electrical parameters. Specifically, the node voltage level can be considered. High-voltage level nodes are typically located at critical hubs in the power grid, connecting long lines with wide coverage. Injecting a test traveling wave at a high-voltage level node allows the traveling wave signal to propagate to various areas of the power grid due to the low-loss characteristics of high-voltage lines, thus effectively detecting faults on long-distance lines. For example, injecting a traveling wave at the bus node of an UHV substation allows the traveling wave signal to be quickly transmitted to inter-regional transmission lines, facilitating the detection of faults on these long-distance lines. Low-voltage level nodes, due to their limited coverage, are generally not used as primary injection points. However, in fault detection in local distribution networks, to accurately locate low-voltage line faults, a traveling wave can be injected at a low-voltage node near the suspected fault area, utilizing the propagation characteristics of low-voltage traveling waves to obtain more accurate fault information.

[0047] In specific embodiments, the injection point of the traveling wave test can be determined based on the node's short-circuit capacity. Specifically, a node with a large short-circuit capacity means it has a strong support capability for the power grid. When a traveling wave is injected at this node, the traveling wave signal can be better maintained and propagated, and is less likely to attenuate too quickly due to factors such as changes in the grid's impedance. This ensures that the traveling wave signal has sufficient strength to reach each measurement end during propagation, improving the accuracy of fault location. For example, near power generation nodes or outgoing nodes of large power plants, these nodes typically have large short-circuit capacities. Injecting traveling waves at these nodes allows the traveling wave signal to propagate more stably in the power grid, ensuring that even if it encounters reflection and refraction at the fault point, a strong echo signal can still be captured by the measurement end.

[0048] In specific embodiments, the injection point can be determined by analyzing node connectivity relationships. Specifically, the topological location of nodes can be analyzed; hub nodes located at the center of the power grid topology and connecting multiple lines are ideal injection point choices. Injecting traveling waves at these nodes allows the waves to propagate simultaneously in multiple directions, covering more lines and areas, thus enabling comprehensive fault detection in the power grid. For example, injecting traveling waves at the junction nodes of a ring power grid or the central substation nodes of a radial power grid allows the waves to propagate along different line branches. By measuring the traveling wave signals returned from each line, the line where the fault is located can be quickly identified. However, for terminal nodes, due to their single connection and limited traveling wave propagation paths, they are generally unsuitable as primary injection points unless there is a specific fault detection requirement for the terminal lines.

[0049] In specific implementations, the influence of adjacent nodes can be considered when selecting the injection point. Specifically, if adjacent nodes have complex power electronic equipment or nonlinear loads, they may interfere with the traveling wave signal. In this case, injecting the traveling wave near these nodes that are significantly affected by interference should be avoided. Instead, nodes with relatively simple and stable adjacent nodes should be selected as injection points to ensure the purity and reliability of the traveling wave signal. For example, in areas with a large number of distributed energy sources and dense power electronic equipment, nodes in these areas should be avoided, and nodes at a certain distance from these areas with relatively stable electrical environments should be selected as injection points to prevent the injected traveling wave signal from being affected by harmonics and other interferences, thus affecting the accuracy of fault location.

[0050] In specific embodiments, the injection point can be determined based on the node's operating status. Specifically, the node load can be monitored in real time, and nodes with lighter loads should be prioritized when determining the injection point. This is because injecting traveling waves into heavily loaded nodes may affect their normal operation and could even cause voltage fluctuations, frequency shifts, and other problems, impacting the stability of the power grid. Nodes with lighter loads, on the other hand, have a stronger capacity to withstand additional traveling wave signals, and the impact of injecting the traveling wave on the power grid's operating status is smaller. For example, during off-peak electricity consumption periods, some industrial nodes have lower loads; selecting these nodes as injection points at this time can effectively detect faults in relevant lines without affecting the normal operation of the power grid. Simultaneously, by dynamically adjusting the selection of injection points through real-time monitoring of node load changes, the safety and effectiveness of the traveling wave injection operation can be ensured.

[0051] In specific implementations, the fault history and risks of nodes can be considered. Nodes that have frequently experienced faults or are located in high-risk areas should be included in the scope of injection points. By injecting traveling waves into these nodes, the operating status of surrounding lines and equipment can be monitored in a focused manner, and potential fault hazards can be detected in a timely manner. For example, in nodes in areas with concentrated old lines or in areas susceptible to natural disasters, regularly injecting traveling waves into these nodes for testing can detect problems such as insulation aging and poor contact in advance, providing a basis for power grid maintenance and repair, and reducing the probability of fault occurrence.

[0052] Furthermore, in some embodiments, step S3, "emitting a test traveling wave from the injection point and acquiring real-time feedback echoes based on pre-set monitoring points," may specifically include: S31. Based on the electrical characteristics and topological location of the nodes, set up corresponding monitoring points; S32. Deploy traveling wave signal acquisition devices at monitoring points to capture the reflection and refraction signals of the test traveling wave in real time; S33. Preprocess the acquired reflected and refracted signals to generate real-time feedback echoes.

[0053] Specifically, in step S3, when the test traveling wave is emitted based on the injection point, the real-time feedback echo is determined in real time according to the pre-set monitoring points. When determining monitoring points in a flexible capacity expansion system, it is necessary to ensure that the traveling wave signal can be accurately captured and the grid status can be fully reflected at the monitoring points. Therefore, monitoring points in a flexible capacity expansion system can be determined based on the electrical characteristics, topological location, and fault risk of the nodes.

[0054] In specific embodiments, monitoring points can be determined based on the electrical characteristics of nodes. Specifically, monitoring points can be determined based on the voltage level of the nodes. High-voltage level nodes are key hubs for power transmission in the power grid, and their operating status is crucial to the stability of the entire power grid. Although traveling waves experience less loss when propagating on high-voltage lines, they are also more susceptible to electromagnetic interference. Setting high-voltage level nodes as monitoring points allows for real-time monitoring of the propagation of traveling waves on long-distance, high-capacity transmission lines, enabling timely detection of potential faults. For example, deploying monitoring equipment at the busbar nodes of UHV substations can effectively capture traveling wave signals and provide early warnings of faults in cross-regional transmission lines. Simultaneously, changes in the voltage, current, and other electrical parameters of these nodes have a significant impact on the power grid, and monitoring their traveling wave signals helps in analyzing the overall operating status of the power grid. Sensitive nodes in flexible capacity expansion systems can also be identified based on key electrical parameters and designated as monitoring points. Specifically, nodes with special electrical parameters, such as nodes with small short-circuit capacity and significant impedance changes, are more susceptible to faults and exhibit significant changes in traveling wave characteristics. Nodes with small short-circuit capacity experience large voltage fluctuations and complex traveling wave reflection and refraction characteristics during faults; nodes with significant impedance changes exhibit altered traveling wave propagation characteristics due to load or line conditions. Designating these nodes as monitoring points allows for rapid fault location by analyzing changes in traveling wave signals. For example, at distributed generation (DG) connection points, fluctuations in power output alter node impedance; setting up monitoring points here allows for timely monitoring of the impact on traveling waves and determination of the DG's connection status with the main grid.

[0055] In specific embodiments, monitoring points can also be determined based on the node topology. Specifically, key nodes can be identified based on the topology of the flexible expansion system and designated as monitoring nodes. Nodes located at the core of the power grid topology and connecting multiple lines, such as junctions in ring grids and substation nodes in radial grids, exhibit complex traveling wave propagation paths. Setting these nodes as monitoring points allows for the acquisition of traveling wave information from multiple directions, facilitating the analysis of traveling wave propagation patterns and the identification of faulty lines. For example, in a ring grid, traveling wave signals monitored at junctions originate from different directions. By comparing the arrival time and amplitude of traveling waves from each direction, the faulty line segment can be quickly identified. Simultaneously, the monitoring data from these key nodes can be used to verify data from other monitoring points, improving the accuracy of fault location. Specifically, line branches and interconnection nodes in the flexible expansion system can also be identified as monitoring nodes. Line branch nodes and interconnection nodes are critical locations for changes in traveling wave propagation direction and energy distribution. Setting monitoring points at these nodes allows for the monitoring of the propagation and reflection of traveling waves between different lines, accurately determining whether a fault occurs on a branch line or interconnection line. For example, in a distribution network, there are numerous branch nodes. Setting these nodes as monitoring points can promptly detect branch line faults, narrow down the fault area, and improve fault location efficiency. Furthermore, interconnecting nodes connect different regional power grids; monitoring their traveling wave signals helps to understand the inter-regional power transmission status and the scope of fault impact.

[0056] In specific embodiments, monitoring points can also be determined based on node failure risk. Specifically, nodes with frequent historical failures can be identified as monitoring nodes. Nodes that have frequently experienced failures in the past may have problems with equipment aging, environmental factors, or operating conditions, and the probability of recurrence is high. Focusing on monitoring these nodes allows for real-time tracking of equipment status changes and early detection of potential failures. For example, if a switch node on a certain line has repeatedly failed due to poor contact, setting up a monitoring point there and monitoring changes in traveling wave signal characteristics can promptly detect abnormalities in the switchgear, such as abnormal traveling wave reflection caused by increased contact resistance, thereby enabling maintenance measures to prevent recurrence. Specifically, nodes in high-risk areas can also be identified as monitoring nodes. Nodes located in special environments such as areas prone to natural disasters are greatly affected by external factors and have a high risk of failure. Setting up monitoring points at nodes in these areas allows for timely understanding of the impact of environmental factors on power grid equipment and assessment of the likelihood of failure. For example, in coastal areas, factors such as typhoons and salt spray can easily damage power equipment. Monitoring nodes in these areas with traveling waves can analyze changes in traveling wave signals to determine whether the insulation performance of the equipment has deteriorated or the line has been damaged, providing data support for power grid disaster prevention and mitigation. In specific embodiments, monitoring nodes can also be determined based on the redundancy and complementarity of the node layout. Specifically, to ensure the reliability of monitoring data and avoid the loss of critical information due to the failure of a single monitoring point, redundant monitoring points are appropriately added near important areas or critical lines. These redundant monitoring points can employ different monitoring technologies or equipment, such as some using traveling wave current-based monitoring devices and others using traveling wave voltage-based monitoring devices, complementing and verifying each other. When the data from a certain monitoring point is abnormal, it can be checked against the data from other redundant monitoring points, improving the accuracy and reliability of fault location. In addition, complementary monitoring points are rationally laid out according to the different regions and functional characteristics of the power grid. Setting up monitoring points on the power source side can monitor the emission and propagation of traveling waves from the power source; setting up monitoring points on the load side can understand the changes of traveling waves under the influence of load; setting up monitoring points at different voltage level transition nodes can analyze the characteristics of traveling waves during voltage transformation. Through this complementary layout, the propagation law of traveling waves in the entire power grid can be comprehensively grasped, and the fault location can be located more accurately. For example, in a flexible expansion system containing multiple distributed power sources and complex loads, comprehensive analysis of monitoring data from the power supply side, load side, and voltage conversion nodes can more clearly determine whether the fault occurred in the power generation stage, the load usage stage, or the voltage conversion stage.

[0057] Furthermore, in some embodiments, step S4, "determining the first failure mode of the flexible expansion system based on the properties of the test traveling wave and nodes using a pre-trained graph neural network model," may specifically include: S41. Merge the test wave and node attributes to generate a node feature vector containing wave characteristics; S42. Construct graph-structured data based on node feature vectors and the topological connectivity of the flexible expansion system; S43. Input the graph structure data into a pre-trained graph neural network model to obtain the hidden feature representations of the nodes; S44. Based on the hidden feature representation of the nodes, output the first fault mode, which is the fault probability distribution of each node in the flexible expansion system.

[0058] Furthermore, in some embodiments, step S41, "fusing the test wave with the node's attributes to generate a node feature vector containing the wave's features," may specifically include: S411. Extract the waveform features of the test traveling wave at multiple monitoring points to obtain a set of traveling wave features, including arrival time, amplitude and polarity; S412. Associate and concatenate the traveling wave feature set with the node attributes to construct the original node feature vector; S413. Denoise and normalize the original node feature vectors to generate node feature vectors containing traveling wave features.

[0059] Specifically, for step S4, based on the test traveling wave and the attributes of any node in the flexible capacity expansion system, and using a pre-trained graph neural network model, the first failure mode of the flexible capacity expansion system is determined, which may include: (1) Data acquisition and preprocessing, including: 1.1 Acquiring Test Traveling Wave Data: After transmitting the test traveling wave at the selected injection point, feedback echo data of the test traveling wave is collected in real time from pre-set monitoring points. This data includes key characteristics such as the arrival time, amplitude, and polarity of the traveling wave. For example, if the monitoring point records the arrival time of the traveling wave as t1, the amplitude as A1, and the polarity as positive, this data will serve as an important basis for subsequent analysis. Since the actual acquired traveling wave data may contain noise interference, filtering algorithms, such as wavelet filtering, are needed to denoise the traveling wave signal and improve data quality.

[0060] 1.2 Collect node attribute data. Collect attribute information for each node in the flexible capacity expansion system, including electrical attributes such as the node's voltage level, load type and size, short-circuit capacity, number and parameters of connected lines, and the node's location information in the power grid topology. For example, a node with a voltage level of 220kV, connected to 3 transmission lines, and a load type mainly of industrial load with a size of 50MW, can reflect the node's operating characteristics and electrical properties in the power grid.

[0061] 1.3 Data Fusion: The collected test traveling wave data is fused with node attribute data. Relevant characteristics of the traveling wave at each node (such as the time of arrival at the node, amplitude changes at the node, etc.) are used as additional attributes of the node. For example, the time t1 of the traveling wave arriving at a node is used as a new attribute of that node, integrated with existing electrical attributes, etc., to form comprehensive data containing both traveling wave characteristics and inherent node attributes, providing comprehensive input information for the graph neural network model.

[0062] (2) Input to the graph neural network model, including: 2.1 Constructing a power grid graph structure: The flexible expansion system is abstracted as a graph structure, where nodes correspond to buses, transformers, distributed power source access points, etc., in the power grid, and edges correspond to connections such as transmission lines. The merged node attribute data is assigned to the corresponding nodes, while the edge attributes are set according to line parameters (such as resistance, reactance, conductance, and susceptance). For example, for a transmission line with resistance R and reactance X, R and X are used as attributes of the corresponding edge, thus constructing a complete power grid graph structure data containing rich information.

[0063] 2.2 Data Input Model: The constructed power grid diagram structure data is input into a pre-trained graph neural network model. After receiving this data, the model propagates and processes information between nodes based on its own network structure and the trained parameters. In the graph convolutional network (GCN), nodes aggregate information from their neighbors and update it in conjunction with their own features. Through multi-layer network computation, a high-level representation containing power grid topology and fault-related features is extracted.

[0064] (3) Model calculation and output, including: 3.1 Model Calculation Process: The graph neural network model performs calculations based on the input power grid graph data, according to the mapping relationships learned during training. In each layer's calculation, potential patterns and rules in the data are gradually mined through node information aggregation and feature transformation operations. For example, in the Graph Attention Network (GAT), by calculating the attention weights between nodes, the model adaptively focuses on the information of nodes and edges important for fault location, strengthening the propagation and processing of key information and suppressing interference from irrelevant information, thereby extracting fault features more accurately.

[0065] 3.2 Output the first fault mode. After multiple layers of calculation by the model, the first fault mode is finally output. The first fault mode is a comprehensive judgment and representation of the current power grid fault state by the model. It describes the possible fault situations in the power grid in a specific form (such as a vector, probability distribution, etc.). For example, the output is a probability vector, where each element of the vector corresponds to a region or node in the power grid, representing the probability of a fault occurring in that region or node. The region or node with the highest probability value is the most likely fault location judged by the model.

[0066] In a specific embodiment, the first fault mode can be a representation of the location and attributes of the fault node.

[0067] Specifically, if the first fault mode is output as a probability distribution, the node or area with the highest probability value is the most likely fault location. For example, in the output probability distribution vector, the probability value corresponding to node A is 0.8, which is much higher than that of other nodes. Therefore, it can be determined that a fault has most likely occurred at the location of node A. In this way, the fault is located to a specific node or line area, providing a clear direction for subsequent fault investigation and repair.

[0068] Graph neural network models fully consider the topology of the power grid when processing data. The output of the first fault mode is closely related to the graph structure. The model determines the location of the fault in the topology based on the connectivity of nodes and the information propagation paths within the graph. For example, if there is an anomaly in the information propagation between the faulty node and several other key nodes, the model will reflect this anomaly in the output fault mode, thereby helping to determine the specific location of the faulty node in the power grid topology. Even if the fault location is in a complex multi-branch network structure, it can be accurately determined through graph structure analysis.

[0069] During the data preprocessing stage, various node attributes (such as electrical and topological attributes) are fused with traveling wave characteristics. The graph neural network model extracts and analyzes these fused attribute features during computation. The first fault mode contains a comprehensive reflection of node attributes. For example, if the voltage level attribute of the faulty node plays a crucial role in fault diagnosis during model calculation, then features related to that voltage level will be reflected in the first fault mode. By analyzing the output results, abnormalities in the faulty node's attributes can be inferred, such as abnormal voltage levels or unreasonable changes in load type and magnitude.

[0070] Different combinations of node attributes correspond to different fault modes. Graph neural network models learn the correlation between attributes and fault modes through training. The first fault mode is actually an abstract representation of the fault situation under a specific combination of attributes. For example, when a node has a small short-circuit capacity and the connected lines are subject to harmonic interference, it will output a specific fault mode. By interpreting the first fault mode and comparing it with the fault modes in the training data, the attribute characteristics of the faulty node can be identified, thereby gaining a deeper understanding of the causes and background of the fault and providing more comprehensive information for fault analysis and handling.

[0071] Furthermore, in some embodiments, step S5, "determining the second fault mode of the flexible capacity expansion system based on real-time feedback echoes," may specifically include: S51. Preprocess and extract features from the real-time feedback echo to obtain the echo feature vector of each monitoring point; S52. Integrate the echo feature vectors of each monitoring point and combine them with the topology of the flexible expansion system to generate a system-level echo feature matrix; S53. Based on the system-level echo characteristic matrix, the second fault mode is determined by analyzing the time difference of arrival of traveling waves and the propagation path. The second fault mode includes the location and type of the fault point.

[0072] Specifically, for step S5, the second failure mode of the flexible capacity expansion system is determined based on the real-time feedback echo. This requires fully utilizing the real-time feedback echo data acquired by the monitoring nodes and conducting in-depth analysis in conjunction with system characteristics.

[0073] During the echo data preprocessing stage, the real-time feedback echo data collected by the monitoring nodes may contain various noises, such as harmonic interference from power electronic equipment and environmental electromagnetic noise. Digital filtering techniques, such as FIR (Finite Impulse Response) filters and IIR (Infinite Impulse Response) filters, can be used. Filtering parameters are designed based on the frequency characteristics of the echo signal to remove noise interference and retain the effective signal components. For sudden impulse noise, a median filtering algorithm can be used to eliminate isolated noise points by replacing each point in the signal sequence with the median value of its neighborhood.

[0074] Key features are extracted from the cleaned data, including traveling wave arrival time, amplitude, polarity, rise time, and fall time. The traveling wave arrival time is crucial for determining fault location and is obtained using a high-precision time measurement device. Amplitude reflects the severity of the fault and the attenuation of the traveling wave during propagation. Polarity helps analyze the reflection and refraction characteristics of the traveling wave. Rise and fall times are related to the fault type and line parameters. For example, for short-circuit faults, the traveling wave amplitude is usually large, and the rise time is short; while for open-circuit faults, the traveling wave characteristics may exhibit specific amplitude changes and polarity reversals.

[0075] Since data collected from different monitoring nodes may differ in amplitude and other aspects, the extracted feature data is normalized to facilitate subsequent analysis and processing. A minimum-maximum normalization method is used to map the data to the [0,1] interval, thereby eliminating the influence of data units and making different feature data comparable.

[0076] In specific embodiments, a corresponding echo feature vector can be constructed based on any cleaned echo data. Specifically, the feature data extracted from each monitoring node can be integrated to form a complete echo feature vector. The feature data of each monitoring node is used as a sub-vector of the vector, arranged according to the node's number or position in the power grid topology. For example, if there are n monitoring nodes in the system, and each node extracts m features, then the dimension of the echo feature vector is n*m. This integration method can comprehensively reflect the traveling wave feature information of each monitoring point in the entire power grid, providing a rich data foundation for subsequent analysis. Specifically, time stamps can be added to the echo feature vector to record the specific time of data acquisition. Time stamps are crucial for analyzing the propagation process of traveling waves and the development of faults. By comparing the echo feature vectors at different times, the changing pattern of traveling wave features over time can be observed, and it can be determined whether a fault is developing or a new fault has occurred. The time stamps use a high-precision timestamp format, accurate to the microsecond level, to ensure the accuracy of the time information.

[0077] In specific embodiments, a historical fault database can also be established to store echo feature vectors and their corresponding fault modes under different fault types, locations, and severity levels. The currently acquired echo feature vector is compared with the data in the historical database, and similarity indices such as Euclidean distance and cosine similarity are calculated. The fault mode corresponding to the historical data with the highest similarity is selected as the preliminary second fault mode prediction. For example, if the current echo feature vector has the smallest Euclidean distance to the feature vector of a short-circuit fault on a certain line in the historical database, the preliminary prediction is that the fault mode may be a short circuit on that line.

[0078] Specifically, machine learning algorithms, such as Support Vector Machines (SVM), Random Forests, and Neural Networks, can be used to classify and recognize patterns in echo feature vectors. First, the algorithm is trained using historical fault data, and its parameters are adjusted to accurately map echo feature vectors to corresponding fault mode categories. After training, real-time echo feature vectors are input into the trained model, which then outputs a predicted second fault mode. For example, a fault mode recognition model can be built using deep neural networks. Through nonlinear transformations and feature extraction of multiple layers of neurons, it can automatically learn the complex mapping relationship between echo features and fault modes, improving the accuracy and efficiency of fault mode recognition.

[0079] In specific embodiments, the grid topology and current operating status of the flexible capacity expansion system can also be considered to verify and correct the initially determined second fault mode. If the fault location corresponding to the initially predicted fault mode does not match the connection relationship of lines and load distribution in the grid topology, the echo feature vector is re-analyzed, and the fault mode prediction is adjusted. For example, if the predicted fault occurs on a certain line, but the line is currently in an unloaded state, and the traveling wave propagation characteristics are different from those under full load, the fault mode needs to be corrected in combination with the line's operating status to consider whether other factors affect the echo characteristics.

[0080] In a specific embodiment, the fault node can be located based on the time difference of arrival of the traveling wave. Specifically, the time difference can be calculated based on the arrival times of the traveling wave recorded by multiple monitoring nodes. Combined with the propagation speed of the traveling wave on different lines, the location of the fault node can be determined using the principle of two-end or multi-end traveling wave ranging. Assuming the propagation speed of the traveling wave on a certain line is known to be v, and the arrival times of the traveling wave recorded by two monitoring nodes A and B are respectively... t A and t B Then the distance from the fault point to node A is... d A It can be calculated using the following formula: Where L is the line length between nodes A and B. The second fault mode includes time difference and calculated distance information. By analyzing this data, the line and approximate location of the faulty node can be accurately determined.

[0081] In a specific embodiment, the second fault mode is also used to record the propagation path information of the traveling wave in the power grid. By analyzing the changes in echo characteristics at different monitoring nodes, the propagation direction and nodes traversed by the traveling wave are inferred. If the amplitude of the traveling wave detected by a monitoring node suddenly increases or its polarity changes, it indicates that the traveling wave may have passed through a nearby fault node. Combining the power grid topology, the propagation path of the traveling wave is traced step by step to finally determine the specific location of the fault node. For example, if the traveling wave starts from the power source and, after passing through a series of nodes, exhibits abnormal characteristics at a certain node, and the traveling wave characteristics of subsequent nodes also conform to the fault propagation law, then that node can be identified as a fault node.

[0082] In specific embodiments, the second fault mode characterizes the fault type and severity of a fault node through features such as amplitude, rise time, and fall time in the echo feature vector. Different fault types, such as short circuits, open circuits, and ground faults, exhibit different traveling wave characteristics. Short circuit faults typically cause a sharp increase in traveling wave amplitude and a short rise time; open circuit faults may cause polarity reversal and specific amplitude changes in the traveling wave; ground faults produce unique zero-sequence component characteristics in the traveling wave. By analyzing these characteristics, the second fault mode can accurately determine the fault type. Simultaneously, the magnitude of the traveling wave amplitude also reflects the severity of the fault; a larger amplitude indicates a potentially more severe fault. Specifically, the electrical attributes of the fault node, such as voltage level, short-circuit capacity, and load type, affect the propagation and reflection characteristics of the traveling wave, and these effects are reflected in the second fault mode. When a high-voltage level node experiences a fault, the traveling wave propagates over a longer distance with relatively less energy attenuation; when a node with a large short-circuit capacity experiences a fault, the traveling wave reflection coefficient is larger, resulting in a higher echo amplitude; and when different load types experience faults, the traveling wave characteristics will vary due to the different impedance characteristics of the load. The second fault mode, by analyzing the relationship between echo characteristics and node electrical attributes, can infer the electrical properties of the faulty node, providing more comprehensive information for fault analysis and handling. For example, if the echo characteristics show that the traveling wave has low attenuation and large amplitude during propagation, combined with the grid node attribute information, it can be inferred that the faulty node may be located at a high voltage level with a large short-circuit capacity.

[0083] Furthermore, in some embodiments, step S6, "determining the fault node of the flexible capacity expansion system based on the first fault mode and the second fault mode," may specifically include: S61. Based on the first fault mode and the second fault mode, generate a fusion decision factor. The fusion decision factor is used to quantify the degree of consistency between the first fault mode and the second fault mode in terms of fault location and attributes. S62. Determine the final failure node of the flexible expansion system based on the fusion decision factors; S63. When the fusion decision factor is lower than the preset threshold, trigger the optimization feedback signal for the test wave injection strategy or graph neural network model.

[0084] Specifically, for step S6, the fault node of the flexible capacity expansion system is determined based on the first fault mode and the second fault mode. Determining the fault node of the flexible capacity expansion system requires comprehensive use of information from the first and second fault modes. Analyzing the differences between the first and second fault modes can improve the accuracy of fault node location.

[0085] In a specific embodiment, the first fault mode uses a graph neural network model to output the probability distribution or location range of the faulty node based on the power grid topology, node attributes, and test traveling wave characteristics. The second fault mode determines the fault location based on the real-time feedback echoes from monitoring nodes, utilizing information such as the time difference of arrival (TOA) and propagation path of the traveling wave. The location information determined by the two modes is fused. If there is an overlap in the fault location range determined by the two modes, then this overlapping area is highly likely to be the location of the faulty node. For example, the first fault mode indicates that the fault occurs in the vicinity of a substation with a probability of 80%; the second fault mode, through the calculation of the TOA, determines that the fault is on a line connected to that substation. If this line is within the area determined by the first fault mode, then the nodes on this line need to be investigated more thoroughly.

[0086] In a specific embodiment, the first fault mode infers the attribute characteristics of the faulty node, such as fault type and abnormal electrical properties, from a comprehensive analysis of the overall network structure and injected traveling waves. The second fault mode obtains clues about the faulty node's attributes from real-time echo characteristics, such as judging the severity and type of the fault by the traveling wave amplitude and rise time. The faulty node attribute information obtained from the two modes is integrated and mutually verified. If the first fault mode determines that the fault is a short circuit and the faulty node has a small short-circuit capacity; and the second fault mode also determines that it is a short circuit fault based on a large echo amplitude and short rise time, and the node's load has recently increased significantly, then it can be concluded that the faulty node is a node with a small short-circuit capacity and a short circuit caused by a sudden load change.

[0087] In a specific embodiment, if the fault node locations determined by the two modes coincide within a certain error range, and the inferred fault type, node attributes, and other information are consistent, then the node can be identified as a fault node. For example, if the location error range is set to 5% of the line length, and the difference between the fault node locations determined by the first fault mode and the second fault mode is less than this error range, and both modes determine the fault to be a ground fault, and the grounding resistance parameters of the involved nodes are abnormal, then the node can be identified as a fault node.

[0088] In specific embodiments, when the results of the two modes differ to some extent but are also partially correlated, a weighted decision-making mechanism is employed. Weights are assigned to the two modes based on historical accuracy data under different conditions. For example, under normal power grid operation, the first fault mode determined by the graph neural network model has higher accuracy and is assigned a weight of 0.6; while at the moment of fault occurrence, the second fault mode based on real-time echoes is more effective for quickly locating the fault and is assigned a weight of 0.4. The fault node is comprehensively determined through weighted calculations, such as calculating the weighted average of the fault node locations determined by the two modes, as the final fault node location.

[0089] In specific embodiments, the data for the first fault mode comes from the power grid model, node attributes, and injected test traveling waves, focusing on analyzing faults from the overall network structure and pre-set detection signals. The second fault mode relies on feedback echoes collected in real time by monitoring nodes, paying more attention to the instantaneous traveling wave characteristics at the time of fault occurrence. If there are differences between the results of the two modes, it is necessary to analyze whether this is due to limitations in data acquisition. For example, the injected test traveling waves may not completely cover certain areas, making the first fault mode's fault judgment inaccurate for those areas; while the second fault mode may also cause result deviations due to unreasonable monitoring node placement failing to collect key traveling wave information. The first fault mode is based on a graph neural network model, predicting faults by learning from a large amount of historical data. Its advantage lies in considering the comprehensive influence of network topology and node attributes, but it may be limited by incomplete training data. The second fault mode uses algorithms such as traveling wave ranging and feature analysis, which react quickly to real-time traveling wave signals, but is susceptible to noise and interference. When analyzing differences, it is necessary to investigate whether the limitations of the algorithms themselves are the cause. In complex electromagnetic interference environments, the extraction of traveling wave features for the second fault mode may be biased; and if the graph neural network model does not fully learn the features of a specific fault scenario, it will also affect the accuracy of the first fault mode.

[0090] In specific embodiments, the data acquisition and processing mechanisms are improved to address differences in data sources. The coverage and density of monitoring nodes are increased to ensure that real-time echo data comprehensively reflects the power grid status. The injection strategy for test traveling waves is optimized, such as adjusting the injection point location and waveform parameters, so that the test traveling waves can better detect various areas of the power grid. Simultaneously, data preprocessing is strengthened to improve data quality and reduce the impact of noise and interference on both modes. For example, more advanced filtering algorithms are used to process real-time echo data, and the input data cleaning process for the graph neural network model is improved.

[0091] In specific embodiments, the algorithms for the two modes are improved based on differences in algorithm principles. For the graph neural network model, the diversity of training data is increased to simulate more complex fault scenarios and improve the model's generalization ability; techniques such as transfer learning are introduced to optimize model parameters using data from other similar power grid systems. For the traveling wave analysis algorithm for the second fault mode, more robust feature extraction methods are studied, such as a deep learning-based traveling wave feature extraction network; the traveling wave ranging algorithm is optimized to consider the impact of more practical factors on traveling wave propagation and improve positioning accuracy.

[0092] In specific embodiments, a dynamic fusion strategy is established to flexibly adjust the weights and fusion methods of the first and second fault modes in the process of determining fault nodes based on factors such as grid operating status and fault type. For example, when the grid load is stable and the operating status is normal, the first fault mode is used as the primary mode and the second fault mode as a secondary mode for verification; when the grid is operating under high load or there are complex interferences, the weight of the second fault mode is increased to enhance the analysis of real-time traveling wave data. Simultaneously, the application effects of the two modes in different scenarios are continuously summarized, and the fusion strategy is continuously optimized to effectively improve the accuracy of fault node determination.

[0093] In summary, the fault location method for flexible expansion systems based on traveling wave injection provided in this embodiment proposes a fault location framework based on the collaboration of traveling wave injection and graph neural networks. Through composite waveform design, multi-terminal synchronous monitoring, and dynamic topology modeling, it achieves high-precision location of fault nodes in complex flexible power grids, thereby improving the accuracy of fault location for flexible expansion systems of power grids and thus improving the efficiency of power grid maintenance.

[0094] It should be understood that, although Figure 2 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 2 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0095] To facilitate better implementation of the fault location method for flexible capacity expansion systems based on traveling wave injection according to the embodiments of this application, this application also provides a fault location system for flexible capacity expansion systems based on traveling wave injection, which is based on the aforementioned fault location method for flexible capacity expansion systems based on traveling wave injection. The meanings of the terms used are the same as in the aforementioned fault location method for flexible capacity expansion systems based on traveling wave injection, and specific implementation details can be found in the descriptions in the method embodiments.

[0096] Please see Figure 3 , Figure 3This is a schematic diagram of the structure of a fault location system for a flexible capacity expansion system based on traveling wave injection provided in an embodiment of this application. Specifically, the fault location system may include a test traveling wave module 201, an injection point module 202, a feedback echo module 203, a first fault module 204, a second fault module 205, and a fault determination module 206, as follows: The test traveling wave module 201 is used to determine the test traveling wave based on the attributes of the nodes in the flexible expansion system; Injection point module 202 is used to determine the injection point of the test traveling wave based on the node's attributes; Feedback echo module 203 is used to emit a test traveling wave from the injection point and acquire real-time feedback echoes based on pre-set monitoring points; The first fault module 204 is used to determine the first fault mode of the flexible expansion system based on the properties of the test traveling wave and nodes, through a pre-trained graph neural network model. The second fault module 205 is used to determine the second fault mode of the flexible expansion system based on the real-time feedback echo. The fault determination module 206 is used to determine the fault node of the flexible expansion system based on the first fault mode and the second fault mode.

[0097] Furthermore, in some embodiments, the test traveling wave module 201 is specifically used for: Obtain the node's attributes, which include at least one of electrical parameters, connection relationships, and operating status; Adjust the frequency and amplitude of the test traveling wave based on the node's attributes; Based on the topological location of the node and its neighboring nodes, design the waveform and encoding method for testing the traveling wave.

[0098] Furthermore, in some embodiments, the test traveling wave module 201 is specifically used for: Based on the node's voltage level and load type, generate initial traveling wave frequency and initial amplitude parameters; Based on the initial traveling wave frequency and initial amplitude parameters, and combined with the real-time operating status of the node, the optimized traveling wave transmission parameters are calculated. Configure and launch a test traveling wave based on optimized traveling wave launch parameters.

[0099] Furthermore, in some embodiments, the injection point module 202 is specifically used for: Based on the electrical parameters of the nodes, nodes with high voltage levels or short-circuit capacity greater than the preset capacity are selected as candidate injection points. Based on the topological location of the node, the hub node that connects multiple lines is selected as the injection point first; The selection of injection points is dynamically adjusted based on the node's operating status and fault history.

[0100] Furthermore, in some embodiments, the feedback echo module 203 is specifically used for: Based on the electrical characteristics and topological location of the nodes, corresponding monitoring points are set; Deploy traveling wave signal acquisition devices at monitoring points to capture the reflection and refraction signals of the test traveling wave in real time; The collected reflected and refracted signals are preprocessed to generate real-time feedback echoes.

[0101] Furthermore, in some embodiments, the first fault module 204 is specifically used for: The test wave and node attributes are fused to generate a node feature vector containing wave characteristics; Graph structure data is constructed based on node feature vectors and the topological connection relationships of the flexible expansion system. Input graph structure data into a pre-trained graph neural network model to obtain the hidden feature representations of nodes; Based on the hidden feature representation of nodes, the first fault mode is output, which is the fault probability distribution of each node in the flexible expansion system.

[0102] Furthermore, in some embodiments, the first fault module 204 is specifically used for: The waveform features of the test traveling wave at multiple monitoring points are extracted to obtain a set of traveling wave features, which include arrival time, amplitude, and polarity. The traveling wave feature set is associated with and concatenated with the node attributes to construct the original node feature vector; The original node feature vectors are denoised and normalized to generate node feature vectors containing traveling wave characteristics.

[0103] Furthermore, in some embodiments, the second fault module 205 is specifically used for: The real-time feedback echoes are preprocessed and feature extracted to obtain the echo feature vectors of each monitoring point; By integrating the echo feature vectors from each monitoring point and combining them with the topology of the flexible expansion system, a system-level echo feature matrix is ​​generated. Based on the system-level echo feature matrix, the second fault mode is determined by analyzing the time difference of arrival of traveling waves and the propagation path. The second fault mode includes the location and type of the fault point.

[0104] Furthermore, in some embodiments, the fault determination module 206 is specifically used for: Based on the first and second failure modes, a fusion decision factor is generated. The fusion decision factor is used to quantify the degree of consistency between the first and second failure modes in terms of failure location and attributes. Based on the fusion decision factors, the final failure node of the flexible expansion system is determined; When the fusion decision factor is lower than a preset threshold, an optimization feedback signal is triggered for the test wave injection strategy or graph neural network model.

[0105] In summary, the flexible capacity expansion system fault location system based on traveling wave injection provided in this embodiment includes: a traveling wave testing module 201 determining the test traveling wave based on the attributes of nodes in the flexible capacity expansion system; an injection point module 202 determining the injection point of the test traveling wave based on the attributes of nodes; a feedback echo module 203 emitting the test traveling wave from the injection point and acquiring real-time feedback echoes based on pre-set monitoring points; a first fault module 204 determining the first fault mode of the flexible capacity expansion system based on the test traveling wave and node attributes using a pre-trained graph neural network model; a second fault module 205 determining the second fault mode of the flexible capacity expansion system based on the real-time feedback echo; and a fault determination module 206 determining the fault node of the flexible capacity expansion system based on the first and second fault modes. This embodiment can improve the accuracy of fault node location in the flexible capacity expansion system, solve the location error problem that may be caused by relying on only a single signal source or a single judgment method, achieve precise location of fault nodes in the flexible capacity expansion system, and thus improve the efficiency of power grid maintenance.

[0106] Specific limitations regarding the fault location system for flexible capacity expansion systems based on traveling wave injection can be found in the limitations of the fault location method for flexible capacity expansion systems based on traveling wave injection mentioned above, and will not be repeated here. Each module in the aforementioned fault location system for flexible capacity expansion systems based on traveling wave injection can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in the computer device, or stored in software in the memory of the computer device, so that the processor can call and execute the corresponding operations of each module.

[0107] Furthermore, embodiments of this application also provide an electronic device, such as... Figure 4 As shown, it illustrates a structural schematic diagram of the electronic device involved in the embodiments of this application, specifically: The electronic device may include components such as a processor 301 with one or more processing cores, a memory 302 with one or more computer-readable storage media, a power supply 303, and an input unit 304. Those skilled in the art will understand that... Figure 4 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein: The processor 301 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines, and performs various functions and processes data by running or executing software programs and / or modules stored in the memory 302, and by calling data stored in the memory 302, thereby providing overall monitoring of the electronic device. Optionally, the processor 301 may include one or more processing cores; preferably, the processor 301 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 301.

[0108] The memory 302 can be used to store software programs and modules. The processor 301 executes various functional applications and fault location methods for flexible expansion systems based on traveling wave injection by running the software programs and modules stored in the memory 302. The memory 302 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 302 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 302 may also include a memory controller to provide the processor 301 with access to the memory 302.

[0109] The electronic device also includes a power supply 303 that supplies power to various components. Preferably, the power supply 303 can be logically connected to the processor 301 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 303 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0110] The electronic device may also include an input unit 304, which can be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0111] Although not shown, the electronic device may also include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 301 in the electronic device loads the executable files corresponding to the processes of one or more applications into the memory 302 according to the following instructions, and the processor 301 runs the applications stored in the memory 302 to realize various functions, as follows: Based on the attributes of nodes in the flexible capacity expansion system, a test traveling wave is determined; based on the attributes of nodes, the injection point of the test traveling wave is determined; the test traveling wave is emitted from the injection point, and real-time feedback echoes are obtained based on pre-set monitoring points; according to the test traveling wave and node attributes, a first fault mode of the flexible capacity expansion system is determined through a pre-trained graph neural network model; based on the real-time feedback echoes, a second fault mode of the flexible capacity expansion system is determined; based on the first fault mode and the second fault mode, the faulty node of the flexible capacity expansion system is determined.

[0112] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0113] The embodiments of this application can improve the accuracy of locating fault nodes in flexible capacity expansion systems, solve the problem of positioning errors that may be caused by relying on a single signal source or a single judgment method, achieve accurate positioning of fault nodes in flexible capacity expansion systems, and thus improve the efficiency of power grid maintenance.

[0114] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0115] Therefore, embodiments of this application provide a storage medium storing multiple instructions that can be loaded by a processor to execute steps in any of the fault location methods for flexible capacity expansion systems based on traveling wave injection provided in embodiments of this application. For example, the instructions can execute the following steps: Based on the attributes of nodes in the flexible capacity expansion system, a test traveling wave is determined; based on the attributes of nodes, the injection point of the test traveling wave is determined; the test traveling wave is emitted from the injection point, and real-time feedback echoes are obtained based on pre-set monitoring points; according to the test traveling wave and node attributes, a first fault mode of the flexible capacity expansion system is determined through a pre-trained graph neural network model; based on the real-time feedback echoes, a second fault mode of the flexible capacity expansion system is determined; based on the first fault mode and the second fault mode, the faulty node of the flexible capacity expansion system is determined.

[0116] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0117] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0118] Since the instructions stored in the storage medium can execute the steps in any of the fault location methods for flexible capacity expansion systems based on traveling wave injection provided in the embodiments of this application, the beneficial effects that any of the fault location methods for flexible capacity expansion systems based on traveling wave injection provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0119] The foregoing has provided a detailed description of a fault location method and system for a flexible capacity expansion system based on traveling wave injection, as provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A fault location method for a flexible capacity expansion system based on traveling wave injection, characterized in that, include: Based on the attributes of nodes in the flexible expansion system, the test traveling wave is determined; Based on the attributes of the node, the injection point of the test traveling wave is determined; The test traveling wave is emitted from the injection point, and real-time feedback echoes are obtained based on pre-set monitoring points; Based on the properties of the test traveling wave and the node, the first failure mode of the flexible expansion system is determined using a pre-trained graph neural network model. Based on the real-time feedback echo, the second fault mode of the flexible capacity expansion system is determined; Based on the first fault mode and the second fault mode, the fault node of the flexible expansion system is determined.

2. The fault location method for a flexible capacity expansion system based on traveling wave injection according to claim 1, characterized in that, The determination of the test traveling wave based on the attributes of nodes in the flexible capacity expansion system includes: Obtain the attributes of the node, including at least one of electrical parameters, connection relationships, and operating status; Based on the attributes of the node, adjust the frequency and amplitude of the test traveling wave; Based on the topological location of the node and its neighboring nodes, design the waveform and encoding method of the test traveling wave.

3. The fault location method for a flexible capacity expansion system based on traveling wave injection according to claim 2, characterized in that, Adjusting the frequency and amplitude of the test traveling wave based on the attributes of the node includes: Based on the voltage level and load type of the node, the initial traveling wave frequency and initial amplitude parameters are generated; Based on the initial traveling wave frequency and the initial amplitude parameters, and in conjunction with the real-time operating status of the node, the optimized traveling wave emission parameters are calculated. Based on the optimized traveling wave transmission parameters, the test traveling wave is configured and transmitted.

4. The fault location method for a flexible capacity expansion system based on traveling wave injection according to claim 1, characterized in that, Determining the injection point of the test traveling wave based on the attributes of the node includes: Based on the electrical parameters of the nodes, nodes with high voltage levels or short-circuit capacity greater than a preset capacity are selected as candidate injection points. Based on the topological location of the node, a hub node connecting multiple lines is preferentially selected as the injection point; The selection of injection points is dynamically adjusted based on the operating status and fault history of the nodes.

5. The fault location method for a flexible capacity expansion system based on traveling wave injection according to claim 1, characterized in that, The step of emitting the test traveling wave from the injection point and obtaining real-time feedback echoes based on pre-set monitoring points includes: Based on the electrical characteristics and topological location of the nodes, corresponding monitoring points are set; A traveling wave signal acquisition device is deployed at the monitoring point to capture the reflection and refraction signals of the test traveling wave in real time; The collected reflected and refracted signals are preprocessed to generate real-time feedback echoes.

6. The fault location method for a flexible capacity expansion system based on traveling wave injection according to claim 1, characterized in that, The step of determining the first failure mode of the flexible capacity expansion system based on the test traveling wave and the attributes of the nodes using a pre-trained graph neural network model includes: The test traveling wave is fused with the attributes of the node to generate a node feature vector containing the traveling wave characteristics; Based on the node feature vectors and the topological connection relationships of the flexible expansion system, a graph structure data is constructed. The graph structure data is input into the pre-trained graph neural network model to obtain the hidden feature representation of the node; Based on the hidden feature representation of the node, the first fault mode is output, which is the fault probability distribution of each node in the flexible expansion system.

7. The fault location method for a flexible capacity expansion system based on traveling wave injection according to claim 6, characterized in that, The step of fusing the test traveling wave with the attributes of the node to generate a node feature vector containing traveling wave features includes: The waveform features of the test traveling wave at multiple monitoring points are extracted to obtain a traveling wave feature set, wherein the waveform features include arrival time, amplitude and polarity; The traveling wave feature set is associated with and concatenated with the attributes of the node to construct the original node feature vector; The original node feature vector is denoised and normalized to generate a node feature vector containing traveling wave characteristics.

8. The fault location method for a flexible capacity expansion system based on traveling wave injection according to claim 1, characterized in that, The step of determining the second fault mode of the flexible capacity expansion system based on the real-time feedback echo includes: The real-time feedback echoes are preprocessed and feature extracted to obtain the echo feature vectors of each monitoring point; The echo feature vectors from each monitoring point are integrated, and combined with the topology of the flexible expansion system, to generate a system-level echo feature matrix. Based on the system-level echo feature matrix, the second fault mode is determined by analyzing the time difference of arrival and propagation path of the traveling wave. The second fault mode includes the location and type of the fault point.

9. The fault location method for a flexible capacity expansion system based on traveling wave injection according to claim 1, characterized in that, The step of determining the fault node of the flexible capacity expansion system based on the first fault mode and the second fault mode includes: Based on the first fault mode and the second fault mode, a fusion decision factor is generated, which is used to quantify the degree of consistency between the first fault mode and the second fault mode in terms of fault location and attributes. Based on the fusion decision factors, the final failure node of the flexible expansion system is determined; When the fusion decision factor is lower than a preset threshold, an optimization feedback signal is triggered for the test wave injection strategy or the graph neural network model.

10. A fault location system for a flexible capacity expansion system based on traveling wave injection, characterized in that, include: The test traveling wave module is used to determine the test traveling wave based on the attributes of nodes in a flexible expansion system; The injection point module is used to determine the injection point of the test traveling wave based on the attributes of the node. The feedback echo module is used to emit the test traveling wave from the injection point and acquire real-time feedback echoes based on pre-set monitoring points; The first fault module is used to determine the first fault mode of the flexible expansion system based on the test traveling wave and the attributes of the node through a pre-trained graph neural network model. The second fault module is used to determine the second fault mode of the flexible expansion system based on the real-time feedback echo. The fault determination module is used to determine the fault node of the flexible expansion system based on the first fault mode and the second fault mode.

Images