Fault positioning and tracing method and device for operation information of medium-low voltage power distribution network

Abstract

The invention provides a fault positioning and tracing method and device for operation information of a medium-low voltage power distribution network, and the method comprises the steps: collecting multi-source heterogeneous data through sensing nodes arranged in the medium-low voltage power distribution network, and carrying out the data processing, and forming model input data; then obtaining a current fault state classification result by using a pre-trained fusion fault detection model; then calculating the delay inequality of fault signal mutation among the sensing nodes according to the classification result, and deducing a preliminary positioning result by combining the physical space position coordinates of the nodes and the delay inequality; mapping the preliminary positioning result into an actual power distribution network topology, and combining with topological structure correction to obtain a corrected positioning result; and finally, based on a fault state classification result and a correction positioning result, matching an event with the most similar cause path from historical fault events to obtain a traceability result. Through multi-source data monitoring and fault tracing, the risk of missed judgment and misjudgment is reduced, systematic tracing is realized, and accurate decision support is provided for operation and maintenance.

Description

Technical Field

[0001] This invention belongs to the field of power system distribution network operation monitoring technology, specifically relating to a method and device for fault location and source tracing of medium- and low-voltage distribution network operation information. Background Technology

[0002] In the development of power systems, the distribution network, as a key link connecting power sources and users, is crucial for ensuring power supply stability and reliability. With the deep integration of distributed power sources, energy storage devices, and diverse loads, the distribution network topology is becoming increasingly complex, and its operating conditions exhibit strong volatility and uncertainty. The probability of various types of faults, such as single-phase grounding, three-phase short circuits, equipment disconnections, and compound faults, has increased significantly, placing higher demands on rapid fault detection, accurate fault location, and causal tracing.

[0003] Currently, various technical approaches have emerged in the field of power distribution network fault monitoring and handling. These mainly include monitoring methods based on single electrical parameters, communication link characteristic detection technology, independent fiber optic sensing monitoring technology, and fault diagnosis schemes based on traditional rule-driven or machine learning models. For example, some existing technologies identify faults by collecting electrical parameters such as voltage, current, and power, or indirectly reflect the state of the power distribution network using communication indicators such as bit error rate and signal strength. Other technologies introduce models such as graph convolutional neural networks to achieve fault mode recognition and location by constructing feature template libraries and dynamic graph networks, and optimize model parameters through real-time data updates to adapt to topology changes. These technologies have met the basic requirements for fault detection and location in specific scenarios, but they have not yet broken through the limitations of the traditional technical framework.

[0004] However, existing technologies still have some problems. On the one hand, single-sensor technologies can only capture fault characteristics in one dimension, which is prone to missed or false faults. On the other hand, existing technologies mostly stop at fault detection and location, and cannot provide operation and maintenance personnel with accurate cause judgment and decision support, making it difficult to meet the fault handling needs of smart distribution networks. Summary of the Invention

[0005] In view of this, the present invention provides a method and apparatus for fault location and source tracing of medium- and low-voltage distribution network operation information. It aims to realize real-time monitoring of the operation status of the distribution network, rapid detection and accurate location of faults based on multi-source sensing data, and systematic source tracing analysis of fault causes, so as to improve the accuracy and efficiency of distribution network fault handling and ensure the safe, stable and reliable operation of the power distribution system.

[0006] To achieve the above objectives, the technical solution provided by the present invention is as follows:

[0007] In a first aspect, the present invention provides a method for fault location and source tracing of medium- and low-voltage distribution network operation information, comprising the following steps:

[0008] Data on the distribution network and the environment are collected by sensing nodes deployed in the medium- and low-voltage distribution network, and multi-source heterogeneous data are formed based on the collected data.

[0009] Data processing is performed on multi-source heterogeneous data to form model input data;

[0010] Based on the model input data, a pre-trained fusion fault detection model is used to detect faults in the distribution network and obtain the current fault state classification results of the distribution network.

[0011] The time delay difference of fault signal abrupt change between sensing nodes is calculated based on the fault state classification results, and the preliminary fault location results are derived by combining the physical spatial coordinates of the sensing nodes and the time delay difference.

[0012] The preliminary location results are mapped to the actual distribution network topology, and the preliminary location results are corrected based on the actual distribution network topology to obtain the corrected location results.

[0013] Based on the fault state classification results and the corrected location results, the fault events with the most similar causal paths are matched from historical fault events, and the source tracing results are obtained based on the matched fault events.

[0014] Furthermore, based on the fault state classification results, the time delay difference of fault signal abrupt changes between sensing nodes is calculated, and combined with the physical spatial coordinates of the sensing nodes and the time delay difference, the preliminary fault location results are derived, including:

[0015] The type of fault signal feature is determined based on the fault state classification results; the fault signal feature is extracted from multi-source heterogeneous data and is used to reflect the characteristics of the fault state of the distribution network.

[0016] Using multi-source heterogeneous data, calculate the time difference of abrupt changes in the characteristics of corresponding fault signals between adjacent sensing nodes;

[0017] Based on the physical spatial coordinates of the sensing nodes and the signal propagation speed, a set of positioning equations based on the signal propagation time difference is constructed.

[0018] By using a set of positioning equations and corresponding time differences among multiple sensing signal nodes, the coordinates of the fault point, which serve as the preliminary positioning result, are obtained.

[0019] Furthermore, the preliminary location results are mapped to the actual distribution network topology, and the preliminary location results are corrected based on the actual distribution network topology to obtain the corrected location results, including:

[0020] Establish a distribution network topology adjacency matrix; the topology adjacency matrix is ​​used to represent the electrical connectivity between each sensing node;

[0021] Project the coordinates of the fault point in the preliminary location results to the nearest topology node or line segment in the distribution network topology search space.

[0022] By using the amplitude of electrical quantity changes of adjacent sensing nodes after projection, the coordinates of the fault point are corrected to obtain the corrected coordinates of the fault point as the corrected location result.

[0023] Furthermore, based on the fault state classification results and the corrected location results, the fault events with the most similar causal paths are matched from historical fault events, and the source tracing results are obtained based on the matched fault events, including:

[0024] Construct a knowledge graph for fault tracing; the knowledge graph contains a set of historical fault entities and the relationships between entities; the entity set includes at least fault type, equipment information, environmental factors, and operating conditions;

[0025] Extract the real-time fault feature vectors corresponding to the fault state classification results and the corrected location results;

[0026] Calculate the similarity between the real-time fault feature vector and the feature vectors of each historical fault entity in the historical fault entity set;

[0027] Calculate the overall credibility score for each possible cause of failure based on similarity, and take the cause of failure with the highest overall credibility score as the matching result;

[0028] Based on the causes of failure, possible evolution paths of failure are deduced, and failure evolution risk paths are obtained;

[0029] The corrected location results, fault causes, and fault evolution risk paths will be used as the source tracing results.

[0030] Furthermore, the multi-source heterogeneous data is processed to form the model input data, including:

[0031] Time synchronization of multi-source heterogeneous data is achieved by uniformly aligning data with different sampling frequencies and communication delays through interpolation methods to obtain the first synchronized data sequence.

[0032] The first synchronization data sequence is denoised to obtain the second synchronization data sequence;

[0033] By removing outliers from the second synchronization data sequence, a third synchronization data sequence is obtained.

[0034] Time-domain features, frequency-domain features, and time-series mutation features are extracted from the third synchronous data sequence to obtain the fault feature vector as the model input data.

[0035] Furthermore, the fault detection model is integrated into a multi-layer neural network model built based on Transformer, LSTM, or GRU.

[0036] Furthermore, the sensing node is configured with multiple types of sensing devices, which include at least:

[0037] Power distribution fiber optic sensors, low-voltage carrier sensors, power wireless sensors, and environmental sensors.

[0038] Secondly, the present invention provides a fault location and source tracing device for medium- and low-voltage distribution network operation information, comprising:

[0039] The data acquisition module is used to collect power distribution network data and environmental data through sensing nodes deployed in the medium- and low-voltage power distribution network, and to form multi-source heterogeneous data based on the collected data;

[0040] The data processing module is used to process multi-source heterogeneous data to form model input data;

[0041] The fault prediction module is used to detect faults in the distribution network based on the model input data and a pre-trained fusion fault detection model, and to obtain the current fault state classification result of the distribution network.

[0042] The preliminary location module is used to calculate the time delay difference of fault signal abrupt change between sensing nodes based on the fault state classification results, and to derive the preliminary location result of the fault by combining the physical spatial coordinates of the sensing nodes and the time delay difference.

[0043] The positioning correction module is used to map the preliminary positioning results to the actual distribution network topology, and to correct the preliminary positioning results based on the actual distribution network topology to obtain the corrected positioning results.

[0044] The fault tracing module is used to match the fault events with the most similar causal paths from historical fault events based on the fault status classification results and the corrected location results, and to obtain the tracing results based on the matched fault events.

[0045] Thirdly, the present invention provides a computer device, the device including a processor and a memory:

[0046] The memory is used to store computer programs and send the instructions of the computer programs to the processor;

[0047] The processor executes, according to the instructions of the computer program, a fault location and source tracing method for medium- and low-voltage power distribution network operation information, as described in the first aspect.

[0048] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a fault location and source tracing method for medium- and low-voltage power distribution network operation information as described in the first aspect.

[0049] In summary, this invention provides a method for fault location and source tracing of medium- and low-voltage distribution network operation information. This method first collects distribution network data and environmental data through sensing nodes deployed in the medium- and low-voltage distribution network, and forms multi-source heterogeneous data based on the collected data. Then, it processes the multi-source heterogeneous data to form model input data. Based on the model input data, a pre-trained fusion fault detection model is used to detect distribution network faults, obtaining the current fault state classification result. The time delay difference of fault signal abrupt changes between sensing nodes is calculated based on the fault state classification result, and the preliminary fault location result is derived by combining the physical spatial coordinates of the sensing nodes and the time delay difference. The preliminary location result is mapped to the actual distribution network topology, and the preliminary location result is corrected based on the actual distribution network topology structure to obtain a corrected location result. Finally, based on the fault state classification result and the corrected location result, the fault event with the most similar causal path is matched from historical fault events, and the source tracing result is obtained based on the matched fault event. This invention, through multi-source data monitoring and fault tracing, not only overcomes the limitations of a single perception dimension and comprehensively depicts fault characteristics to significantly reduce the risk of missed or misjudged faults, but also achieves systematic tracing by establishing the correlation between fault status, location results, and historical cause paths, providing precise decision support for operation and maintenance, significantly improving the efficiency and reliability of power distribution network fault handling, and ensuring the safe and stable operation of the power distribution system.

[0050] The present invention also provides a fault location and source tracing device, computer equipment and computer-readable storage medium for medium- and low-voltage distribution network operation information, which have similar effects to the above methods when implemented, and will not be described in detail here. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart illustrating a method for fault location and source tracing of medium- and low-voltage distribution network operation information provided in an embodiment of the present invention;

[0053] Figure 2 This is a block diagram of a fault location and tracing device for medium- and low-voltage power distribution network operation information provided in an embodiment of the present invention;

[0054] Figure 3 This is a block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0055] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0056] Please see Figure 1 This embodiment provides a method for fault location and source tracing of medium- and low-voltage distribution network operation information, including the following steps:

[0057] S11: Collect distribution network data and environmental data through sensing nodes arranged in the medium-low voltage distribution network, and form multi-source heterogeneous data based on the collected data.

[0058] It should be noted that sensing nodes are data acquisition terminals deployed on power distribution network lines, equipment, or in the surrounding environment, and have the function of sensing and transmitting physical quantities such as voltage, current, and temperature.

[0059] Distribution network data directly reflects the operating status of the distribution network, including line voltage, current, power, equipment temperature, and switch status.

[0060] Environmental data refers to external environmental parameters that affect the operation of the power distribution network, including ambient temperature, humidity, rainfall, wind speed, and icing thickness.

[0061] Multi-source heterogeneous data is a mixed collection of data from different acquisition sources (different types of sensing nodes) and with different data formats / types (such as numerical, state, and time series).

[0062] In this step, sensing nodes collect core data on the operation of the distribution network and data from the surrounding environment at a preset frequency, and then aggregate the dispersed data through a communication network. Due to differences in data sources (different node functions) and data types (such as current being a continuous value, and switch status being a discrete state), the aggregated data forms multi-source heterogeneous data.

[0063] S12: Process multi-source heterogeneous data to form model input data.

[0064] It should be noted that data processing is a set of operations for processing multi-source heterogeneous data, including data preprocessing and feature extraction. The purpose of preprocessing is to remove outliers, missing values, and redundant information from the multi-source heterogeneous data, correct distortion caused by equipment errors and transmission interference during data acquisition, and standardize heterogeneous data of different formats (such as numerical, state, and time-series) and magnitudes to eliminate compatibility issues caused by differences in data sources, ensuring data integrity, consistency, and reliability. The purpose of feature extraction is to extract key features strongly correlated with the fault state of the distribution network (such as current mutation characteristics, voltage time-series trend characteristics, and environmental factor correlation characteristics) to form a feature set for model prediction.

[0065] S13: Based on the model input data, use the pre-trained fusion fault detection model to detect faults in the distribution network and obtain the current fault status classification results of the distribution network.

[0066] It should be noted that the fusion fault detection model is a composite model that integrates the advantages of multiple detection algorithms (such as machine learning algorithms, deep learning algorithms, and traditional thresholding algorithms). This model can output the distribution network operation status determination result (i.e., the current fault status classification result of the distribution network), including specific classifications such as "no fault", "short circuit fault", "ground fault", and "overload fault".

[0067] S14: Calculate the time delay difference of fault signal abrupt change between sensing nodes based on the fault state classification results, and deduce the preliminary fault location results by combining the physical spatial coordinates of the sensing nodes and the time delay difference.

[0068] It should be noted that fault signal abrupt change refers to the sudden change in physical quantities of the distribution network generated at the moment a fault occurs (such as during a short circuit or grounding). Common manifestations include sudden increase in current, sudden drop in voltage, and sudden change in power.

[0069] The time delay difference refers to the time difference between different sensing nodes in sensing the sudden change of the fault signal, which reflects the time difference from the fault point to different sensing nodes.

[0070] Physical spatial location coordinates refer to the location identifiers of sensing nodes in the actual geographical environment or power distribution network layout.

[0071] In this step, a sudden signal (such as a sudden current change or a sudden voltage drop) will be generated after a fault occurs. The time it takes for this signal to propagate to surrounding sensing nodes is positively correlated with the distance from the node to the fault point. After confirming the fault occurrence based on the fault state classification results, the time delay difference between the sudden signal received by different sensing nodes is calculated. Combined with the known physical spatial coordinates of the nodes, the signal propagation process is simulated to infer the approximate location of the fault point and obtain a preliminary location result.

[0072] S15: Map the preliminary location results to the actual distribution network topology, and correct the preliminary location results based on the actual distribution network topology to obtain the corrected location results.

[0073] It should be noted that the distribution network topology is the connection relationship and structural layout between lines, equipment and nodes in the distribution network, including the actual physical connection information such as line routes, branching situations and equipment installation locations.

[0074] In this step, the preliminary location result is derived solely from geographical coordinates and does not consider the actual topological features of the distribution network, such as line routes and branch distribution (e.g., the fault point may fall in a non-line area). The preliminary location result is mapped onto the actual distribution network topology map. Based on the actual line path, branch node locations, and other topological information, the preliminary location result is corrected, invalid locations that do not conform to the power grid structure are eliminated, and the specific line section or equipment where the fault point is located is pinpointed, thus obtaining a more accurate corrected location result.

[0075] S16: Based on the fault state classification results and the corrected location results, match the fault events with the most similar causal paths from historical fault events, and obtain the source tracing results based on the matched fault events.

[0076] It should be noted that historical fault events are records of past power distribution network faults stored in the system, including complete information such as fault type, fault location, occurrence time, environmental conditions, cause analysis, and handling results.

[0077] The causal path is the chain of causes that leads to a failure, including the transmission path of direct causes (such as lightning strikes and equipment aging) and indirect influencing factors (such as long-term overload and environmental corrosion).

[0078] In this step, based on the fault status classification results (fault type) and the corrected location results (fault location), historical cases with the closest causal paths are selected from the historical fault event database using similarity algorithms (such as feature matching and path similarity calculation). Referring to the causal analysis results of the matched cases and combining them with the actual data of the current fault (such as environmental conditions and operating parameters), the source tracing results of the current fault are derived, clarifying the root cause of the fault.

[0079] This embodiment provides a method for fault location and source tracing of medium- and low-voltage distribution network operation information. This method, through multi-source data monitoring and fault source tracing, not only makes up for the limitations of a single perception dimension and comprehensively portrays fault characteristics to significantly reduce the risk of missed or misjudged cases, but also achieves systematic source tracing by establishing the correlation between fault status, location results and historical cause paths, providing accurate decision support for operation and maintenance, significantly improving the efficiency and reliability of distribution network fault handling, and ensuring the safe and stable operation of the distribution system.

[0080] In one embodiment of the present invention, the sensing node is configured with multiple types of sensing devices, including but not limited to:

[0081] (1) Distribution fiber optic sensor: Using distributed fiber optic sensing technology (such as DTS (Distributed Temperature Sensing), DAS (Distributed Acoustic Sensing), FBG (Fiber Bragg Grating) strain sensing, etc.), the temperature change, mechanical stress, vibration anomaly and external impact of the cable are monitored in real time along the power distribution line, which helps to identify the fault risk caused by external physical interference.

[0082] (2) Low-voltage carrier sensor: The voltage, current, power, frequency and phase of the node are collected in real time through the power line carrier communication terminal (PLC (Power Line Communication) terminal) to reflect the load changes, power flow fluctuations and short-circuit characteristics of the distribution network.

[0083] (3) Power wireless sensors: Wireless sensing terminals (such as ZigBee (wireless communication technology), LoRa (long-range wireless communication technology), NB-IoT (narrowband Internet of Things)) are deployed in some substations, switching stations, ring network cabinets and branch nodes to obtain communication link quality parameters in real time, and at the same time make up for the operation status sensing capability of some communication dead zones.

[0084] (4) Environmental sensors: Deploy environmental monitoring equipment such as temperature, humidity, vibration, and electromagnetic interference to provide auxiliary background information and facilitate the analysis of the correlation between environmental factors and equipment failures.

[0085] The raw signals collected by various sensors can be formally expressed as follows:

[0086] Electrical quantities:

[0087]

[0088] in, Indicates voltage. Represents current. These represent active power and reactive power, respectively. Indicates the phase angle. Indicates frequency.

[0089] Communication quality:

[0090]

[0091] in, Indicates signal strength. Indicates bit error rate. Indicates packet loss rate. Indicates communication delay.

[0092] Environmental quantity:

[0093]

[0094] in, For temperature, For humidity, For vibration intensity, This refers to the intensity of electromagnetic interference.

[0095] This ultimately constitutes a complete multi-dimensional time-series perceptual feature vector (i.e., multi-source heterogeneous data):

[0096]

[0097] in, Let be the multidimensional time series sensing feature vector at time t.

[0098] In this embodiment, by integrating multiple sensing technologies such as power distribution fiber optic sensing, low-voltage carrier source-load-storage sensing, power wireless sensing, and environmental monitoring, it is possible to acquire multi-dimensional, multi-scale, and multi-level electrical parameters, communication parameters, and environmental information during the operation of the power distribution network in real time, thus overcoming the problem of data one-sidedness that exists in traditional single electrical quantities or single monitoring methods.

[0099] After completing the multi-source sensing data acquisition, to ensure the accuracy, consistency, and usability of the data, it is necessary to preprocess the acquired raw data and extract key features reflecting fault characteristics. In one embodiment of the present invention, multi-source heterogeneous data is processed to form model input data, including:

[0100] S21: Time synchronization of multi-source heterogeneous data is performed by uniformly aligning data with different sampling frequencies and communication delays through interpolation methods to obtain the first synchronized data sequence.

[0101] For example, time synchronization is performed on data from different sensors. Since various sensors may have differences in sampling frequency, communication latency, etc., interpolation methods are needed to uniformly align data from different time scales to obtain a complete and continuous synchronized data sequence. The formula is expressed as:

[0102]

[0103] in, The multidimensional time series sensing feature vector at time t has undergone interpolation processing.

[0104] S22: Denoise the first synchronization data sequence to obtain the second synchronization data sequence.

[0105] For example, various denoising methods can be used to eliminate random noise and environmental interference in synchronized data sequences. For instance, algorithms such as moving average filtering and Kalman filtering can be used to smooth data curves and preserve effective trend characteristics.

[0106] S23: Remove outliers from the second synchronization data sequence to obtain the third synchronization data sequence.

[0107] For example, in the data cleaning stage, a standardized anomaly detection method is introduced to remove potential outliers and isolated data points. The Z-score standardization algorithm is used to calculate the deviation of the standard deviation for each sample point:

[0108]

[0109] in, Represents the sample mean. This represents the sample standard deviation. For samples that satisfy... Data points that are identified as outliers are removed to ensure data quality.

[0110] S24: Extract time-domain features, frequency-domain features, and time-series mutation features from the third synchronous data sequence to obtain the fault feature vector as the model input data.

[0111] After data preprocessing, key features are extracted from the preprocessed multi-source data. Feature extraction includes, but is not limited to:

[0112] Time-domain characteristics, such as mean, maximum, minimum, variance, kurtosis, slope, and rate of change, reflect the direct fluctuation of measured values ​​over time.

[0113] Frequency domain characteristics: Spectral changes are analyzed using Fast Fourier Transform (FFT) to extract frequency domain anomalies such as harmonic amplitude, dominant frequency components, and frequency shift;

[0114] Temporal mutation characteristics: The cumulative sum control chart algorithm is used to detect the location of sequence mutations and help identify the sudden fault points.

[0115] A unified feature vector is formed through feature extraction:

[0116]

[0117] This feature vector serves as the input to the subsequent fault detection and localization model, providing the system with a stable and reliable description of fault features.

[0118] This embodiment ensures the information fusion capability of multi-source heterogeneous data at a unified time scale, while extracting highly relevant fault indication information.

[0119] After feature extraction, an intelligent detection model capable of accurately identifying various fault states in medium- and low-voltage distribution networks needs to be constructed. Considering the complexity of the distribution network operating environment and the diversity of data features, a fusion deep learning model is adopted, combining time-series modeling capabilities and multi-dimensional feature representation capabilities to achieve accurate classification and identification of fault states. In one embodiment of this invention, the fusion fault detection model is a multi-layer neural network model built based on Transformer, LSTM (Long Short-Term Memory) or GRU (Gated Recurrent Unit).

[0120] Specifically, the detection model used in this embodiment is a multi-layer neural network model built based on the Transformer structure or other sequence modeling networks (such as LSTM, GRU). This model can effectively handle the temporal dependence and spatial correlation in the input feature vector and identify pattern changes before and after the fault occurs. The specific form is as follows:

[0121]

[0122] in, The input feature vector set represents the multidimensional perceptual features of each node at the current time. This represents the neural network model obtained after training; y represents the fault classification result output by the model, corresponding to different fault type labels (such as: normal, single-phase grounding, three-phase short circuit, equipment disconnection, communication abnormality, etc.).

[0123] The training objective is to minimize the loss function between the predicted values ​​and the true fault labels. To optimize model parameters θ, commonly used loss functions include the cross-entropy loss function:

[0124]

[0125] Where C represents the number of fault type categories; This is a real label; This represents the failure probability distribution predicted by the model.

[0126] To improve the model's detection capability in situations with few samples and imbalanced classes, Focal Loss can also be introduced as a loss function:

[0127]

[0128] in, γ represents the class weight; γ is the adjustment factor (usually around 2); Focal Loss can improve the model's ability to identify rare fault samples.

[0129] To enhance the model's generalization ability, data augmentation mechanisms can be introduced, including window shifting, amplitude perturbation, and noise injection, to construct more training samples. Furthermore, transfer learning or federated learning can be used to improve the model's adaptability to different power distribution network environments. Ultimately, this fault detection model can achieve online inference in actual operation, performing real-time analysis of the perceived feature sequences and outputting the current fault status.

[0130] In this embodiment, by introducing a deep learning fusion model, the spatiotemporal correlation patterns among multi-source sensing features are fully explored, significantly improving the ability to identify complex fault modes (such as multi-point faults, compound faults, and sudden faults). Compared with the traditional static threshold determination method, it can effectively adapt to various operating conditions and complex load conditions, and the fault identification accuracy is significantly improved.

[0131] After identifying anomalies in the distribution network using a fusion fault detection model, this invention further designs a fault location and tracing method to quickly determine the fault location, narrow down the fault area, and guide subsequent operation and maintenance. This method fully integrates the distribution network topology, the spatiotemporal characteristics of multi-source sensing data, and the node distribution characteristics, effectively improving fault location accuracy. The following describes the fault location and tracing method with some examples.

[0132] In one embodiment of the present invention, the time delay difference of fault signal abrupt change between sensing nodes is calculated based on the fault state classification result, and the preliminary fault location result is derived by combining the physical spatial coordinates of the sensing nodes and the time delay difference, including:

[0133] S31: Determine the type of fault signal features based on the fault state classification results; fault signal features are extracted from multi-source heterogeneous data and are used to reflect the characteristics of the fault state of the distribution network.

[0134] Different fault states are characterized by different types of fault signal features. For example, if the fault state is a short circuit fault, the corresponding fault signal feature type may be a sudden increase in three-phase current or a sudden drop in bus voltage; if the fault state is a ground fault, the corresponding fault signal feature type may be a sudden increase in zero-sequence current, etc.

[0135] S32: Calculate the time difference of the abrupt change in the fault signal characteristics of the corresponding type between adjacent sensing nodes using multi-source heterogeneous data.

[0136] By synchronously collecting signal abrupt change information at multiple sensing nodes, the time difference between the occurrence times of signal abrupt changes at each node can be calculated:

[0137]

[0138] in, , These represent the moments when sensing node i and node j detect the sudden change in the fault signal characteristics determined in step S31, respectively. This reflects the propagation delay of fault signals between different nodes.

[0139] S33: Based on the physical spatial coordinates of the sensing node and the signal propagation speed, construct a set of positioning equations based on the signal propagation time difference.

[0140] Based on the known physical spatial coordinates (x) of the sensing node i ,y i ), combined with signal propagation speed Construct a set of positioning equations based on the propagation time difference:

[0141]

[0142] in, The location of the fault point to be determined; This is the initial time when the fault signal is generated.

[0143] S34: Using the positioning equations and corresponding time differences between multiple sets of sensing signal nodes, the coordinates of the fault point are solved as the preliminary positioning result.

[0144] Since the time difference provided by the aforementioned steps gives the time difference in fault perception between different sensing nodes, combined with the localization equation set, by combining data from multiple nodes, the least squares method or multi-objective optimization algorithm can be used to solve for the fault location coordinates and obtain preliminary localization results.

[0145] In this embodiment, by combining multi-node delay difference calculation with distribution network topology association algorithm, joint modeling of fault signal propagation path and physical network structure is achieved, enabling accurate fault location within seconds. The location error is reduced by more than 50% compared to traditional methods, especially in distribution network scenarios with many branches, dense nodes, and complex topologies.

[0146] Considering the complex branching topology of power distribution networks, simple geometric positioning results may not fall within the actual power path. Therefore, in one embodiment of this invention, a topology constraint correction algorithm is introduced. This involves mapping the initial positioning result to the actual power distribution network topology and correcting the initial positioning result based on the actual power distribution network topology to obtain a corrected positioning result, including:

[0147] S41: Establish the distribution network topology adjacency matrix; the topology adjacency matrix is ​​used to represent the electrical connectivity between each sensing node.

[0148] The adjacency matrix A of the power distribution network topology represents the electrical connectivity between nodes.

[0149] S42: Project the coordinates of the fault point in the preliminary location results to the nearest topology node or line segment in the distribution network topology search space.

[0150] Specifically, the preliminary location results are projected onto the nearest topologically valid node or line segment.

[0151] S43: Using the amplitude of electrical quantity changes of adjacent sensing nodes after projection, the coordinates of the fault point are corrected to obtain the corrected coordinates of the fault point as the corrected location result.

[0152] Specifically, it utilizes the amplitude of electrical quantity changes in neighboring nodes. The weighted adjustment is performed using the following formula:

[0153]

[0154] in, Represents the topologically valid search space; This represents the sensing node adjacent to this location; λ is a weighting factor used to balance the combined effects of geometric distance and neighborhood anomaly amplitude. This represents the magnitude of the electrical quantity change at sensing node k; This represents the set of sensing nodes adjacent to the initially located position (x, y).

[0155] The above steps are used to obtain the coordinates of the fault location, and then determine the power supply circuit and branch number to which it belongs; the scope of the fault is estimated (e.g., the length of the fault section, the number of transformers involved, etc.).

[0156] The fault location method proposed in this embodiment can accurately locate the fault within seconds after it is detected, and the location error can be controlled within tens of meters, which greatly improves the efficiency of fault handling and significantly shortens the power outage time.

[0157] After fault location is completed, this invention further proposes a fault source tracing analysis method, which aims to trace the possible causes, triggers, and evolution paths of faults based on multi-source sensing data and the operating history of the distribution network, providing a basis for decision-making in subsequent operation and maintenance. This invention adopts a source tracing analysis framework based on knowledge graph modeling and feature matching algorithms, making full use of distribution network equipment information, historical fault records, environmental factors, and real-time sensing features to achieve intelligent judgment of fault causes.

[0158] In one embodiment of the present invention, based on the fault state classification result and the corrected location result, the fault event with the most similar causal path is matched from historical fault events, and the source tracing result is obtained based on the matched fault event, including:

[0159] S51: Construct a knowledge graph for fault tracing; the knowledge graph contains a set of historical fault entities and the relationships between entities; the entity set includes at least fault type, equipment information, environmental factors and operating conditions.

[0160] By collecting and organizing historical operation and maintenance data of the distribution network, a fault tracing knowledge graph G=(E,R) is established, where:

[0161] E represents the entity set, including but not limited to: fault types (such as short circuit, grounding, open circuit, insulation failure, mechanical damage, environmental interference, etc.); equipment information (such as switches, circuit breakers, transformers, cables, conductors, etc.); environmental factors (such as temperature and humidity, lightning strikes, vibration, external damage, etc.); and operating conditions (such as power flow fluctuations, overload, load imbalance, etc.). R represents the relationships between entities, describing the possible causal or evolutionary connections between them.

[0162] In knowledge graphs, entities and relationships are modeled using multi-dimensional feature attributes, for example:

[0163]

[0164] in, This represents the set of feature attributes of the i-th entity. , , , , These represent attributes such as type, location, severity, historical frequency of occurrence, and environmental conditions.

[0165] S52: Extract the real-time fault feature vector corresponding to the fault state classification result and the corrected location result.

[0166] When a new fault event is detected, the fault feature vector F is extracted based on the real-time collected data.

[0167] S53: Calculate the similarity between the real-time fault feature vector and the feature vectors of each historical fault entity in the historical fault entity set.

[0168] For example, the formula for calculating feature similarity is as follows:

[0169]

[0170] in, This represents a historical fault entity in the knowledge graph. This indicates the real-time characteristics of the actual fault. Indicate feature similarity measurement functions (such as Euclidean distance, cosine similarity, Mahalanobis distance, etc.); This represents the weighting factor for each feature dimension, used to adjust the contribution of each feature dimension to the similarity score.

[0171] S54: Calculate the overall credibility score for each possible cause of failure based on similarity, and take the cause of failure with the highest overall credibility score as the matching result.

[0172] Based on the similarity score results, calculate the overall credibility score for each possible cause of failure:

[0173]

[0174] Based on the scores, the system outputs a ranking list of causes and their probabilities for maintenance personnel to refer to, helping to accurately pinpoint the root cause of the fault.

[0175] S55: Based on the causes of failure, deduce the possible evolution paths of the failure and obtain the failure evolution risk path.

[0176] For example, based on the knowledge graph, possible evolution paths of faults can be further tracked, and the evolution risk path can be dynamically calculated in combination with the current system state, such as:

[0177]

[0178] in, Indicates from Evolved to Characteristic changes during the process; Indicates the strength of the correlation between the two; This indicates the statistical characteristics of the time intervals in which this evolutionary path occurs in history.

[0179] S56: The corrected location results, fault causes, and fault evolution risk paths will be used as the source tracing results.

[0180] Finally, through the above steps, we obtain a list of the most likely causes of the current failure and their probabilities; a comparison of similar historical failure cases; possible evolution trends and subsequent risk warnings; and operation and maintenance suggestions and intervention measures for reference.

[0181] The above-mentioned source tracing analysis method can effectively improve the ability to diagnose the causes of faults in complex fault scenarios, reduce the fault recurrence rate, and support the formulation of intelligent preventive maintenance and precise operation and maintenance strategies for power grid systems.

[0182] In this embodiment, a knowledge graph of distribution network faults is constructed, integrating multi-source information such as equipment history, environmental factors, and operation and maintenance records to support automated tracing and correlation analysis of fault causes. Based on real-time fault characteristics, it can intelligently recommend possible causes and typical evolution paths, helping to prevent potential risks in advance and improving the scientific and targeted nature of distribution network operation and maintenance.

[0183] The method proposed in this invention is adaptable to medium and low voltage distribution network systems of different scales and structures, and is suitable for various application scenarios such as urban distribution networks, rural distribution networks, industrial park private networks, and microgrids. It can effectively improve the safety, reliability, and intelligence level of power system operation. Furthermore, the algorithm of this invention has a moderate computational load, supports collaborative deployment of edge computing and cloud platforms, and can realize real-time online monitoring and rapid decision support, possessing good engineering feasibility and application promotion value.

[0184] In smart distribution networks, the multi-source communication and sensing fusion algorithm proposed in this invention can be applied to achieve real-time monitoring of the distribution network's operating status, accurate fault location, and intelligent source tracing analysis, thereby improving the safety and operating efficiency of the distribution system.

[0185] First, various types of sensing terminals are deployed on-site, including distribution fiber optic sensors, low-voltage carrier sensing modules, power wireless sensing devices, and environmental status monitoring devices. The sensors collect multi-dimensional data from various nodes of the distribution network in real time, covering voltage amplitude, current changes, power flow fluctuations, equipment status, and external interference information. The sensing data is then aggregated and transmitted to the central monitoring and control platform using low-voltage carrier and wireless communication methods.

[0186] During the system initialization phase, the central control platform sets initial monitoring parameters based on the current distribution network topology and historical operational data, including the activation status of sensing nodes, data acquisition frequency, synchronization time window, and initial model weights. The system automatically configures a feature extraction module to extract multi-dimensional feature vectors, including time domain, frequency domain, and sequence mutation data, providing data support for subsequent intelligent analysis.

[0187] During data processing, the system continuously runs a fault detection model trained on a deep neural network to identify faults in the real-time input feature vector sequence. When the model identifies an abnormal pattern, the system activates the fault location module to calculate the signal response time difference of each monitoring node in real time. Combining the distribution network topology path and the characteristic change information of neighboring nodes, it executes a multi-step correction algorithm to quickly determine the location of the fault.

[0188] After the fault location results are generated, the system further calls the built-in knowledge graph tracing and analysis module. This module assesses potential fault causes and possible causal paths based on the feature matching degree between real-time fault characteristics and existing knowledge entities in the knowledge graph, and generates a fault cause ranking list and early warning report, providing operation and maintenance personnel with accurate handling basis.

[0189] During system operation, the central control platform continuously updates model weights and knowledge graph association strength based on real-time data feedback, gradually improving model adaptability and source tracing analysis accuracy. Simultaneously, the system dynamically optimizes fault detection sensitivity and data acquisition priority based on fault response status and load change trends, effectively balancing monitoring accuracy and communication load.

[0190] Through the above implementation methods, the system can achieve continuous monitoring of complex operating conditions of the distribution network, rapid fault location and intelligent cause analysis, thereby improving the self-healing capability and operational stability of the distribution network.

[0191] Based on the same inventive concept, this application also provides a fault location and tracing device for medium- and low-voltage distribution network operation information, used to implement the fault location and tracing method for medium- and low-voltage distribution network operation information as described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations in the embodiments of the fault location and tracing device for medium- and low-voltage distribution network operation information provided below can be found in the limitations of the fault location and tracing method for medium- and low-voltage distribution network operation information described above, and will not be repeated here.

[0192] Please see Figure 2 This invention also provides a fault location and tracing device for medium- and low-voltage distribution network operation information, comprising:

[0193] The data acquisition module is used to collect power distribution network data and environmental data through sensing nodes deployed in the medium- and low-voltage power distribution network, and to form multi-source heterogeneous data based on the collected data;

[0194] The data processing module is used to process multi-source heterogeneous data to form model input data;

[0195] The fault prediction module is used to detect faults in the distribution network based on the model input data and a pre-trained fusion fault detection model, and to obtain the current fault state classification result of the distribution network.

[0196] The preliminary location module is used to calculate the time delay difference of fault signal abrupt change between sensing nodes based on the fault state classification results, and to derive the preliminary location result of the fault by combining the physical spatial coordinates of the sensing nodes and the time delay difference.

[0197] The positioning correction module is used to map the preliminary positioning results to the actual distribution network topology, and to correct the preliminary positioning results based on the actual distribution network topology to obtain the corrected positioning results.

[0198] The fault tracing module is used to match the fault events with the most similar causal paths from historical fault events based on the fault status classification results and the corrected location results, and to obtain the tracing results based on the matched fault events.

[0199] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0200] Reference Figure 3 The present invention also provides a computer device, including: a memory and a processor and a computer program stored in the memory, wherein when the computer program is executed on the processor, it implements the fault location and source tracing method for medium- and low-voltage distribution network operation information as described in any of the above methods.

[0201] The computer device may be a desktop computer, laptop, handheld computer, or cloud server, etc. This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that... Figure 3 The examples of computer devices are merely examples and do not constitute a limitation on computer devices. They may include more or fewer components than shown in the illustration, or combinations of certain components, or different components. For example, they may also include input / output devices, network access devices, etc.

[0202] The processor referred to can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0203] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard drive or RAM. In other embodiments, the memory may be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory may include both internal and external storage units of the computer device. The memory is used to store the operating system, applications, boot loader, data, and other programs, such as the program code of the computer program. The memory can also be used to temporarily store data that has been output or will be output.

[0204] This invention also provides a computer-readable storage medium storing a computer program thereon. When the computer program is run by a processor, it implements the fault location and source tracing method for medium- and low-voltage distribution network operation information as described in any of the above methods.

[0205] In this embodiment, if the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0206] This invention provides a computer program product, including a computer program that, when executed by a processor, implements a fault location and source tracing method for medium- and low-voltage distribution network operation information as described in any of the above methods.

[0207] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0208] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0209] In the embodiments disclosed in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0210] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for fault location and source tracing of medium- and low-voltage distribution network operation information, characterized in that, Includes the following steps: Data on the distribution network and the environment are collected by sensing nodes deployed in the medium- and low-voltage distribution network, and multi-source heterogeneous data are formed based on the collected data. The multi-source heterogeneous data is processed to form model input data; Based on the input data of the model, a pre-trained fusion fault detection model is used to detect faults in the distribution network and obtain the current fault state classification results of the distribution network. The time delay difference of fault signal abrupt change between sensing nodes is calculated based on the fault state classification results, and the preliminary fault location results are derived by combining the physical spatial coordinates of the sensing nodes and the time delay difference. The preliminary positioning results are mapped to the actual distribution network topology, and the preliminary positioning results are corrected based on the actual distribution network topology to obtain the corrected positioning results; Based on the fault state classification results and the corrected location results, the fault events with the most similar causal paths are matched from historical fault events, and the source tracing results are obtained based on the matched fault events.

2. The fault location and source tracing method for medium- and low-voltage distribution network operation information according to claim 1, characterized in that, Based on the fault state classification results, the time delay difference of fault signal abrupt changes between sensing nodes is calculated, and the preliminary fault location results are derived by combining the physical spatial coordinates of the sensing nodes and the time delay difference, including: The type of fault signal feature is determined based on the fault state classification result; the fault signal feature is extracted from the multi-source heterogeneous data and is used to reflect the fault state of the distribution network. Using the aforementioned multi-source heterogeneous data, the time difference of abrupt changes in the characteristics of corresponding fault signals between adjacent sensing nodes is calculated; Based on the physical spatial coordinates of the sensing nodes and the signal propagation speed, a set of positioning equations based on the signal propagation time difference is constructed. By using the set of positioning equations and the corresponding time differences among multiple sets of sensing signal nodes, the coordinates of the fault point, which serve as the preliminary positioning result, are obtained.

3. The fault location and source tracing method for medium- and low-voltage distribution network operation information according to claim 1, characterized in that, The preliminary location results are mapped to the actual distribution network topology, and the preliminary location results are corrected based on the actual distribution network topology to obtain corrected location results, including: Establish a distribution network topology adjacency matrix; the topology adjacency matrix is ​​used to represent the electrical connectivity between each sensing node; Project the coordinates of the fault point in the preliminary location results onto the nearest topology node or line segment in the distribution network topology search space. The fault point coordinates are corrected by using the electrical quantity change amplitude of the adjacent sensing nodes after projection, and the corrected fault point coordinates are obtained as the corrected positioning result.

4. The fault location and source tracing method for medium- and low-voltage distribution network operation information according to claim 1, characterized in that, Based on the fault state classification results and the corrected location results, the fault events with the most similar causal paths are matched from historical fault events, and the source tracing results are obtained based on the matched fault events, including: Construct a knowledge graph for fault tracing; the knowledge graph contains a set of historical fault entities and the relationships between entities; the set of entities includes at least fault type, equipment information, environmental factors, and operating conditions; Extract the real-time fault feature vectors corresponding to the fault state classification results and the corrected localization results; Calculate the similarity between the real-time fault feature vector and the feature vectors of each historical fault entity in the historical fault entity set; Calculate a comprehensive credibility score for each possible cause of failure based on the similarity, and take the cause of failure with the highest comprehensive credibility score as the matching result; Based on the causes of the failure, possible evolution paths of the failure are deduced, and failure evolution risk paths are obtained; The corrected location result, the cause of the fault, and the risk path of the fault evolution are used as the source tracing result.

5. The fault location and source tracing method for medium- and low-voltage distribution network operation information according to claim 1, characterized in that, The multi-source heterogeneous data is processed to form model input data, including: Time synchronization is performed on the multi-source heterogeneous data. Data with different sampling frequencies and different communication delays are uniformly aligned by interpolation methods to obtain the first synchronized data sequence. The first synchronization data sequence is denoised to obtain the second synchronization data sequence; By removing outliers from the second synchronization data sequence, a third synchronization data sequence is obtained. Time-domain features, frequency-domain features, and time-series mutation features are extracted from the third synchronous data sequence to obtain the fault feature vector that serves as the input data for the model.

6. The fault location and source tracing method for medium- and low-voltage distribution network operation information according to claim 1, characterized in that, The fusion fault detection model is a multi-layer neural network model built based on Transformer, LSTM, or GRU.

7. The method for fault location and source tracing of medium- and low-voltage distribution network operation information according to claim 1 or 2, characterized in that, The sensing node is configured with multiple types of sensing devices, and the multiple types of sensing devices include at least: Power distribution fiber optic sensors, low-voltage carrier sensors, power wireless sensors, and environmental sensors.

8. A fault location and source tracing device for medium- and low-voltage distribution network operation information, characterized in that, include: The data acquisition module is used to collect power distribution network data and environmental data through sensing nodes deployed in the medium- and low-voltage power distribution network, and to form multi-source heterogeneous data based on the collected data; The data processing module is used to process the multi-source heterogeneous data to form model input data; The fault prediction module is used to perform distribution network fault detection based on the model input data and a pre-trained fusion fault detection model to obtain the current fault state classification result of the distribution network. The preliminary location module is used to calculate the time delay difference of fault signal abrupt change between sensing nodes based on the fault state classification results, and to deduce the preliminary location result of the fault by combining the physical spatial coordinates of the sensing nodes and the time delay difference. The positioning correction module is used to map the preliminary positioning result to the actual distribution network topology, and to correct the preliminary positioning result in combination with the actual distribution network topology to obtain the corrected positioning result; The fault tracing module is used to match the fault event with the most similar causal path from historical fault events based on the fault state classification result and the corrected location result, and obtain the tracing result based on the matched fault event.

9. A computer device, characterized in that, The device includes a processor and a memory: The memory is used to store computer programs and send the instructions of the computer programs to the processor; The processor executes, according to the instructions of the computer program, a method for fault location and source tracing of medium- and low-voltage distribution network operation information as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements a fault location and source tracing method for medium- and low-voltage distribution network operation information as described in any one of claims 1-7.

Images