Method and device for detecting feeder node in power distribution network and electronic equipment

Abstract

The embodiment of the invention provides a method and device for detecting feeder nodes in a power distribution network and electronic equipment. The method comprises the following steps: acquiring multi-source node data of each feeder node in a power distribution network within a preset duration; based on the multi-source node data of each feeder node and the topological structure of the power distribution network, performing iterative diffusion processing to obtain an offset feature of each feeder node; generating an offset fingerprint map according to the offset characteristic of each feeder node and the topological structure of the power distribution network; carrying out risk calculation according to the offset fingerprint map to obtain a risk score of each feeder node; and determining a feeder node maintenance queue according to the risk score of each feeder node. The method is used for achieving the effects of accurately positioning a fault source and a fault diffusion path and improving the emergency repair success rate.

Description

Technical Field

[0001] This application relates to the field of power grid fault detection, and in particular to a detection method, device and electronic equipment for feeder nodes in a distribution network. Background Technology

[0002] In a power distribution network, power supply equipment is interconnected through dense lines, forming a complex power transmission network. When a piece of equipment experiences local anomalies such as voltage fluctuations, abnormal temperatures, or acoustic distortion due to a sudden increase in load or environmental disturbances, these anomalies may spread along the lines to adjacent equipment through electrical coupling or thermal conduction, eventually triggering a cascading failure and causing a large-scale power outage.

[0003] Existing operation and maintenance systems mostly rely on single-point monitoring and static scheduling, making it difficult to capture the propagation path and dynamic evolution of anomalies, resulting in delayed fault location and low repair success rate. Summary of the Invention

[0004] This application provides a method, device, and electronic equipment for detecting feeder nodes in a distribution network, which can accurately locate the fault source and fault propagation path, thereby improving the success rate of emergency repairs.

[0005] In a first aspect, embodiments of this application provide a method for detecting feeder nodes in a distribution network, comprising:

[0006] Acquire multi-source node data for each feeder node in the distribution network over a preset time period; perform iterative diffusion processing based on the multi-source node data of each feeder node and the topology of the distribution network to obtain the offset characteristics of each feeder node; generate an offset fingerprint map based on the offset characteristics of each feeder node and the topology of the distribution network; perform risk estimation based on the offset fingerprint map to obtain the risk score of each feeder node; determine the feeder node maintenance queue based on the risk score of each feeder node.

[0007] Secondly, embodiments of this application provide a detection device for feeder nodes in a distribution network, comprising:

[0008] The acquisition module is used to acquire multi-source node data for each feeder node in the distribution network for a preset time period;

[0009] The iterative diffusion module is used to perform iterative diffusion processing based on the multi-source node data of each feeder node and the topology of the distribution network to obtain the offset characteristics of each feeder node.

[0010] The offset fingerprint generation module is used to generate an offset fingerprint based on the offset characteristics of each feeder node and the topology of the distribution network.

[0011] The risk scoring determination module is used to calculate the risk based on the offset fingerprint map and obtain the risk score for each feeder node.

[0012] The maintenance queue determination module is used to determine the maintenance queue of each feeder node based on the risk score of each feeder node.

[0013] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0015] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0016] The detection method, device, and electronic equipment for feeder nodes in a distribution network provided in this application embodiment utilize multi-source node data of each feeder node in the distribution network over a preset time period, along with the distribution network topology, to perform iterative diffusion processing to obtain the offset characteristics of each feeder node. An offset fingerprint map is generated based on these offset characteristics, quantifying the spread of abnormal situations between feeder nodes. Risk estimation is performed based on the offset fingerprint map to obtain a risk score for each feeder node, thereby generating a feeder node maintenance queue. This queue not only reflects high-risk fault sources but also the spread path of faults between feeder nodes. Emergency repairs are performed by referring to the feeder node maintenance queue, improving the success rate of emergency repairs, optimizing the allocation of emergency repair resources, and ensuring the stable operation of the distribution network. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] Figure 1 A schematic diagram of the detection method for feeder nodes in the distribution network provided in this application. Figure 1 ;

[0019] Figure 2 A schematic diagram of the detection method for feeder nodes in the distribution network provided in this application. Figure 2 ;

[0020] Figure 3 A schematic diagram of the detection method for feeder nodes in the distribution network provided in this application. Figure 3 ;

[0021] Figure 4 This is a schematic diagram of the structure of the detection device for feeder nodes in the distribution network provided in this application;

[0022] Figure 5 A schematic diagram of the structure of the electronic device provided in this application.

[0023] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0024] 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 numbers 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 apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0025] The application scenario of the detection method for feeder nodes in the distribution network provided in this application is as follows: In the distribution network, power supply equipment is interconnected through dense lines to form a complex power transmission network. When a device experiences local anomalies such as voltage fluctuations, abnormal temperatures, or acoustic distortion due to a sudden increase in load or environmental disturbances, these anomalies may spread along the line to adjacent devices through electrical coupling or thermal conduction effects, eventually triggering a cascading failure and causing a large-scale power outage.

[0026] Existing operation and maintenance systems largely rely on single-point monitoring and static scheduling, making it difficult to capture the propagation path and dynamic evolution patterns of anomalies. This leads to delayed fault location, unbalanced allocation of emergency repair resources, and prevents maintenance personnel from developing targeted repair strategies before faults spread. Therefore, there is an urgent need for a systematic solution that can integrate multi-dimensional heterogeneous data, determine anomaly propagation, and optimize repair sequences.

[0027] The core technical concept of this application is to quantify the spread of abnormal situations between feeder nodes by unifying the processing of multi-source data and mapping the topology in the distribution network, performing risk estimation based on the offset fingerprint map of the feeder nodes, determining the risk score of each feeder node, and thus obtaining the feeder node maintenance queue. This feeder node maintenance queue can not only reflect high-risk fault sources, but also reflect the spread path of faults between feeder nodes. By referring to the feeder node maintenance queue to perform emergency repair work, the success rate of emergency repair is improved, the allocation of emergency repair resources is optimized, and the stable operation of the distribution network is ensured.

[0028] Figure 1A flowchart illustrating the detection method for feeder nodes in the distribution network provided in this application. Figure 1 The detection method for feeder nodes in a distribution network can be applied to electronic devices, which can be... Figure 1 The terminal or server in the middle; such as Figure 1 As shown, the detection methods for feeder nodes in a distribution network include:

[0029] S101. Synchronously acquire multi-source node data of each feeder node in the distribution network for a preset time period.

[0030] The specific value of the preset duration can be set according to actual needs, and this application embodiment does not limit it; the multi-source node data includes parameters of the feeder node in multiple dimensions.

[0031] The multi-source node data includes: voltage value sequences, current value sequences, temperature change sequences, acoustic signature sequences, and meteorological parameter sequences.

[0032] Specifically, for each feeder node in the distribution network, the node parameters of each feeder node in the distribution network are collected in real time through the sensors of the feeder node. Voltage value sequence, current value sequence, temperature change sequence, and acoustic value sequence are extracted from the sensors of the feeder node, and meteorological parameter series are obtained from the weather station or integrated sensor.

[0033] The voltage value sequence includes: voltage values ​​measured at multiple consecutive time points within a preset time period; current value sequence, including current values ​​measured at multiple consecutive time points within a preset time period; temperature change sequence, including node temperatures measured at multiple consecutive time points within a preset time period; and acoustic value sequence, representing sound pressure levels measured at multiple consecutive time points within a preset time period.

[0034] In practical applications, voltage value sequences, current value sequences, temperature change sequences, and acoustic signature sequences can be acquired using the original sampling frequency to ensure coverage of the entire process of abnormal fluctuations, thereby capturing potential abnormal propagation between feeder nodes in the initial stage of a fault.

[0035] The meteorological parameter sequences include, but are not limited to: ambient temperature offset sequence, humidity sequence, and wind speed sequence.

[0036] Optionally, time point alignment processing is performed on the voltage value sequence, current value sequence, temperature change sequence, voiceprint value sequence, and meteorological parameter sequence; specifically, the timestamps of the voltage value sequence, current value sequence, temperature change sequence, voiceprint value sequence, and meteorological parameter sequence are determined respectively, and the time points of the voltage value sequence, current value sequence, temperature change sequence, voiceprint value sequence, and meteorological parameter sequence are unified to the standard time grid through linear interpolation.

[0037] The standard time grid is defined as a time grid with fixed intervals. If a global time axis is divided into segments (e.g., 1 second), and there are missing values ​​at a certain time point in the voltage value sequence, current value sequence, temperature change sequence, voiceprint value sequence, or meteorological parameter sequence, the parameters at that time point are supplemented by linear interpolation.

[0038] For example, the voltage value is missing at time point t, and interpolation is achieved by formula (1).

[0039] Formula (1): ;

[0040] in, It is the interpolated voltage value at time point t in the voltage value sequence. It is a point in time. The voltage value collected at the location, It is a time point in the voltage value sequence. The voltage value collected at the location; It is a point in time. arrive The interval length.

[0041] For example, the current value is missing at time point t, and interpolation is achieved through formula (2).

[0042] Formula (2): ;

[0043] in, It is the interpolated current value at time point t in the current value sequence. It is a point in time. The current value collected at the location, It is a time point in the current value sequence. The current value collected at the location; It is a point in time. arrive The interval length.

[0044] For example, the node temperature value is missing at time point t, and interpolation is achieved through formula (3).

[0045] Formula (3): ;

[0046] in, It is the nodal temperature value interpolated at time point t. In the sequence of node temperature values, time point The node temperature values ​​collected at the location, In the node temperature value sequence, time point The node temperature values ​​collected at the location; It is a point in time. arrive The interval length.

[0047] For example, the voiceprint value is missing at time point t, and interpolation is achieved through formula (4).

[0048] Formula (4): ;

[0049] in, It is the interpolated voiceprint value at time point t. It is the time point in the voiceprint value sequence. The voiceprint value collected at the location, It is the time point in the voiceprint value sequence. The voiceprint value collected at the location; It is a point in time. arrive The interval length.

[0050] For example, meteorological parameters are missing at time point t, and interpolation is achieved through formula (5).

[0051] Formula (5): ;

[0052] in, These are meteorological parameters interpolated at time point t. It is a time point in the meteorological parameter series. Meteorological parameters collected at the location, It is a time point in the meteorological parameter series. Meteorological parameters collected at the location; It is a point in time. arrive The interval length.

[0053] Linear interpolation ensures the integrity of voltage, current, temperature, acoustic signature, and meteorological parameter sequences, and aligns them across all time points, thus preventing time-series deviations from affecting subsequent anomaly propagation analysis.

[0054] The time-aligned voltage, current, temperature change, acoustic signature, and meteorological parameter sequences are integrated into a multi-source data stream and then subjected to structured transformation to obtain multi-source structured data for the feeder nodes. For example, the multi-source structured data includes multiple key-value records. In each key-value record, the key is the time point after time alignment, and the value is the corresponding voltage, current, temperature change, acoustic signature, and meteorological parameter. Through structured transformation, the raw data is converted into an operable data format, allowing for direct reading and processing of the voltage, current, temperature change, acoustic signature, and meteorological parameter sequences to support real-time quantification of anomaly propagation.

[0055] Through the above process, a time-consistent multi-dimensional input is provided for the quantification of anomaly propagation in the distribution network. In scenarios where increased electricity demand or weather changes cause equipment anomalies, multi-source node data can be directly processed to avoid the impact of time-series deviations on subsequent anomaly propagation analysis. Under the complex topology of the distribution network, the close connection between power supply equipment makes it easy for abnormal signals to propagate along the lines, affecting overall stability. By acquiring and processing multi-source node data in real time, comprehensive input data is provided for subsequent processes, ensuring the accuracy of anomaly propagation quantification and risk scoring.

[0056] S102. Based on the multi-source node data of each feeder node and the topology of the distribution network, perform iterative diffusion processing to obtain the offset characteristics of each feeder node.

[0057] The topology of the distribution network includes all feeder nodes in the distribution network and the lines between the feeder nodes; the connection relationship between the lines between the feeder nodes in the distribution network can be determined through the topology.

[0058] Iterative diffusion processing refers to the process of performing multiple diffusion iterations according to the topological structure.

[0059] Offset characteristics can represent the degree to which a feeder node is affected by the diffusion of other feeder nodes, and can also represent the offset changes that occur in the feeder node itself during the diffusion process.

[0060] Specifically, the multi-source node data of each feeder node is used as the node attribute of each feeder node in the topology of the distribution network. For each feeder node, the attributes of adjacent nodes are integrated into the node attributes of the feeder node through iterative diffusion to obtain the diffused node attributes of the feeder node. Based on the node attributes of the feeder node and the diffused node attributes, the offset characteristics of the feeder node are determined.

[0061] Optionally, based on the node attributes of the feeder node and the attributes of the diffused node, the adjacent fingerprint vector and the self-difference vector are determined, and the adjacent fingerprint vector and the self-difference vector are concatenated to obtain the offset feature; wherein, the adjacent fingerprint vector represents the propagation influence of the adjacent node on the current feeder node, and the self-difference vector represents the deviation of the current feeder node in the diffusion process.

[0062] S103. Generate an offset fingerprint map based on the offset characteristics of each feeder node and the topology of the distribution network.

[0063] Specifically, the offset features are mapped to the topology of the distribution network to obtain the offset fingerprint map.

[0064] Optionally, the offset feature is obtained by concatenating the adjacent fingerprint vector and its own difference vector. The offset feature is directly associated with the corresponding node in the distribution network topology to form an offset fingerprint map. The offset fingerprint map stores the offset features (adjacent fingerprint vector and its own difference vector) of each feeder node in the form of graph node attributes, thereby visualizing the offset pattern of the entire network, which is convenient for capturing the gradient change of anomalies along the conductor direction, so as to determine the distribution and propagation pattern of anomalies in the network topology.

[0065] It should be noted that the close connection between power supply equipment makes it easy for abnormal signals to propagate along the line, affecting the overall stability. However, by converting multi-source node data into offset fingerprints through topology mapping, the dynamic relationship between equipment can be revealed, enabling accurate quantification of abnormal propagation paths and dynamic optimization of maintenance sequences.

[0066] S104. Risk estimation is performed based on the offset fingerprint map to obtain the risk score for each feeder node.

[0067] Specifically, the offset features of each feeder node in the offset fingerprint map include adjacent fingerprint vectors and their own difference vectors. The adjacent fingerprint vectors and their own difference vectors are determined based on multi-source node data. Therefore, the offset features include the offset of multi-source node data for each feeder node. When the multi-source node data includes voltage value sequences, current value sequences, temperature change sequences, acoustic fingerprint value sequences, and meteorological parameter sequences, the offset features include voltage offset sequences, current offset sequences, temperature offset sequences, acoustic fingerprint offset sequences, and meteorological parameter offset sequences.

[0068] Based on the offset characteristics of each feeder node in the offset fingerprint map, the voltage and current differential sequence, temperature offset sequence, and acoustic offset sequence are determined. Risk is estimated based on the voltage and current differential sequence, temperature offset sequence, and acoustic offset sequence to obtain the risk score of each feeder node.

[0069] S105. Determine the feeder node maintenance queue based on the risk score of each feeder node.

[0070] Specifically, multiple feeder nodes are arranged in descending order of risk score to obtain a feeder node maintenance queue.

[0071] It should be noted that the feeder node at the top of the feeder node maintenance queue is the node with the highest risk score. In scenarios where increased power demand or weather changes cause equipment malfunctions, the feeder node maintenance queue can be used to identify the feeder node most prone to risk and the risk propagation path.

[0072] The method for detecting feeder nodes in a distribution network provided in this application uses multi-source node data of each feeder node in the distribution network over a preset time period, along with the topology of the distribution network, to perform iterative diffusion processing to obtain the offset characteristics of each feeder node. An offset fingerprint map is generated based on these offset characteristics, quantifying the spread of abnormal situations between feeder nodes. Risk estimation is performed based on the offset fingerprint map to obtain a risk score for each feeder node, resulting in a feeder node maintenance queue. This queue not only reflects high-risk fault sources but also the spread path of faults between feeder nodes. Emergency repairs are performed by referring to the feeder node maintenance queue, improving the success rate of emergency repairs, optimizing the allocation of emergency repair resources, and ensuring the stable operation of the distribution network.

[0073] In some embodiments, based on the multi-source node data of each feeder node and the topology of the distribution network, an iterative diffusion process is performed to obtain the offset characteristics of each feeder node, including: determining the initial multi-source vector of each feeder node based on the multi-source node data of each feeder node; determining the adjacent nodes of each feeder node based on the topology of the distribution network; performing iterative diffusion process based on the initial multi-source vector of each feeder node and the initial multi-source vector of the adjacent nodes to obtain the diffused multi-source vector of each feeder node; and determining the offset characteristics of each diffused node based on the initial multi-source vector and the diffused multi-source vector of each feeder node.

[0074] Adjacent nodes are other feeder nodes that have line links with feeder nodes.

[0075] Specifically, the voltage value sequence, current value sequence, temperature change sequence, acoustic value sequence, and meteorological parameter sequence of each feeder node are combined to obtain the initial multi-source vector of each feeder node;

[0076] For each feeder node, in the topology of the distribution network, the adjacent nodes of each feeder node are determined. Based on the initial multi-source vector of the feeder node and the initial multi-source vector of the adjacent nodes, iterative diffusion processing is performed to obtain the diffused multi-source vector of the feeder node; as shown in formula (6).

[0077] Formula (6): ;

[0078] in, It is a feeder node In the iteration step The diffusion vector at that location, This represents the diffusion factor (ranging from 0 to 1). Indicates feeder node Adjacent nodes In the iteration step The sum of the diffusion vectors at that point, Indicates feeder node The degree (number of adjacent nodes), It is the initial multi-source vector of the feeder node; after performing a preset number of iterations, the diffused multi-source vector of the feeder node is obtained.

[0079] It should be noted that each feeder node in the distribution network topology performs iterative diffusion processing synchronously to obtain the diffusion multi-source vector of each feeder node.

[0080] In one alternative approach, for each feeder node, the difference between the initial multi-source vector and the diffused multi-source vector of the feeder node is used as the offset feature of each feeder node.

[0081] In the above embodiments, the offset characteristics of each feeder node are determined through iterative diffusion processing, so that the offset characteristics can reflect the diffusion influence of adjacent nodes on each feeder node, realizing the quantification of abnormal propagation, and providing a necessary basis for subsequent calculation of the fault source and the diffusion path of the fault between feeder nodes based on the offset characteristics.

[0082] In one alternative approach, the offset features of each diffusion node are determined based on the initial multi-source vector and diffusion multi-source vector of each feeder node, including: determining the adjacent fingerprint vector corresponding to the diffusion process based on the initial multi-source vector and diffusion multi-source vector of each feeder node; determining the self-difference vector based on the initial multi-source vector and diffusion multi-source vector of each feeder node; and determining the offset features of each diffusion node based on the adjacent fingerprint vector and self-difference vector of each feeder node.

[0083] Among them, the adjacent fingerprint vector is the cumulative value obtained by propagation through adjacent nodes in the multi-source diffusion vector; the self-difference vector represents the difference between the feeder nodes before and after diffusion.

[0084] Specifically, the contribution of adjacent nodes is separated from the diffusion multi-source vector and used as the adjacent fingerprint vector; for example, the adjacent fingerprint vector is determined by formula (7).

[0085] Formula (7): ;

[0086] in, It is the adjacent fingerprint vector. It is the total number of iterations. It is the diffusion factor. , indicating the first During iterative diffusion, the contribution of the initial multi-source vector normalization from adjacent nodes.

[0087] Formula (7) captures the fingerprint influence of adjacent nodes on the current feeder node, with dimensions consistent with the initial multi-source vector, thereby highlighting the potential diffusion path of anomalies between adjacent devices.

[0088] Subtract the initial multi-source vector from the diffused multi-source vector to obtain the self-difference vector, as shown in formula (8).

[0089] Formula (8): ;

[0090] in Indicates feeder node The self-difference vector, Indicates feeder node In the iteration step The obtained diffusion multi-source vector, It is a feeder node The initial multi-source vector is obtained. This difference operation quantifies the difference between local anomalies and adjacent effects, with the dimension of the vector, thereby identifying the local offset intensity of anomaly propagation.

[0091] The offset feature of each feeder node is obtained by combining the adjacent fingerprint vector of each feeder node with its own difference vector.

[0092] In the above embodiments, the influence of adjacent nodes propagating to feeder nodes is quantified by the adjacent fingerprint vector, and the offset features are determined by the adjacent fingerprint vector and its own difference vector, thereby improving the accuracy of the offset features.

[0093] In some embodiments, risk estimation is performed based on the offset fingerprint to obtain a risk score for each feeder node, including: determining the voltage-current difference sequence, temperature offset sequence, and acoustic offset sequence for each feeder node based on the offset fingerprint; determining the fluctuation spike amount based on the voltage-current difference sequence; determining the temperature-acoustic offset rate based on the temperature offset sequence and the acoustic offset sequence; performing risk estimation using an enhancement tree model based on the fluctuation spike amount and the temperature-acoustic offset rate to obtain the diffusion coefficient for each feeder node; and determining the risk score for each feeder node based on the diffusion coefficient for each feeder node.

[0094] Specifically, based on the offset characteristics of each feeder node in the offset fingerprint map, voltage offset sequence, current offset sequence, acoustic offset sequence, and temperature offset sequence are obtained.

[0095] The voltage offset sequence includes voltage offset features at multiple time points, and the current offset sequence includes current offset features at multiple time points. Subtracting the voltage offset features and current offset features at the same time point yields the voltage-current difference sequence. The voltage-current difference sequence can reflect the transient imbalance between electrical parameters and reflect the fluctuation signal caused by the anomaly, so that subsequent peak detection can focus on the propagation signs of imbalance drive and facilitate the reflection of abnormal electrical correlation between equipment.

[0096] Based on the calculated voltage-current difference sequence, the upper and lower envelopes are extracted using the rolling median envelope. The rolling median envelope is generated by smoothing the voltage-current difference sequence by taking the median within a fixed window. Peaks are identified between the upper and lower envelopes, where a peak is defined as a local peak exceeding the envelope threshold. This is determined by scanning the voltage-current difference sequence to find the local maximum value and comparing its difference with the envelope value. Then, the area of ​​each peak is integrally calculated, where the area is obtained by summing the absolute differences between the sequence values ​​from the peak's start point to its end point and the envelope value. The area of ​​each peak is then normalized by dividing the area of ​​each peak by the total length of the voltage-current difference sequence. Finally, the normalized areas of all peaks are summed to obtain the fluctuation peak quantity. This calculation quantifies the cumulative intensity of peaks, highlighting the prominence of abnormal fluctuations, thus enabling fluctuation characteristics to be more accurately integrated into the propagation risk assessment and facilitating the capture of the impact of transient events on network stability.

[0097] For the acoustic offset sequence and temperature offset sequence, dynamic time registration is applied to align them. Dynamic time registration achieves nonlinear alignment by constructing a distance matrix and finding the path with the minimum cumulative cost. First, the difference between aligned points on the path is calculated. Then, these differences are squared, averaged, and the square root is taken to obtain the root mean. Finally, the root mean is divided by the overall mean of the temperature offset sequence, where the overall mean is the sum of all point values ​​in the temperature offset sequence divided by the number of points, to obtain the temperature-acoustic offset deviation rate. This calculation measures the deviation ratio between temperature and acoustic signals, reflecting the synchronization deviation of thermoacoustic anomalies, thereby capturing environmentally induced propagation factors and facilitating the transformation of multimodal deviations into a unified anomaly index.

[0098] The diffusion coefficient is obtained by feeding the fluctuation peak quantity and the temperature-sound deviation rate into the enhancement tree model.

[0099] The diffusion coefficient of each feeder node can be determined based on the above process. After obtaining the diffusion coefficient of each feeder node, the risk gradient of the node path corresponding to each feeder node can be determined based on the diffusion coefficient of each feeder node, and then the risk score of each feeder node can be determined.

[0100] Among them, the boosting tree model, referring to gradient boosting decision trees, is an ensemble learning algorithm that gradually improves the overall performance of the model by sequentially building multiple weak learners. The following sections will introduce the model's boosting mechanism, construction process, training and optimization methods, and parameter settings.

[0101] The enhancement mechanism of the boosting tree model is based on the gradient boosting principle, which combines multiple weak learners into a strong learner through iteration. Specifically, this mechanism employs a forward step-by-step addition strategy, starting from an initial model and adding a new tree at each step. This tree fits the negative gradient of the residuals of the preceding model to minimize the overall loss function. This process is similar to gradient descent optimization, where each tree represents a step in the optimization direction, ensuring that the model gradually approaches the optimal solution, thereby capturing nonlinear relationships and reducing bias and variance in complex data patterns.

[0102] The construction process of the augmented tree model is carried out sequentially. First, an initial model is built, which is the mean or log odds of the target variable. Second, the residuals of the current model are calculated, which are the differences between the actual and predicted values. Then, a decision tree is fitted to each residual. This tree builds nodes by recursively splitting the feature space until a stopping condition (such as maximum depth or minimum number of samples) is met. The new tree is multiplied by the learning rate and added to the current model. This process is repeated until the preset number of trees is reached or the loss converges, thus forming the final ensemble model. The prediction of this model is the weighted sum of the outputs of all trees. For example, in a regression task, if the initial model predicts a constant value of 5, while the actual value is 8, then the residual is 3. Subsequent trees are fitted to this residual to gradually correct the prediction.

[0103] The training and optimization of augmented tree models are achieved by minimizing a loss function such as mean squared error or log loss. During training, the algorithm uses a variant of gradient descent: for each iteration, the loss gradient of the current model on the training samples is calculated, and the decision tree is then fitted to that negative gradient as an approximate optimization step. For optimization, regularization terms (such as leaf node weight penalties) are introduced to prevent overfitting, and subsampling (randomly selecting a subset of samples) or column sampling (randomly selecting features) is employed to enhance generalization ability. The overall training process is iterative until the validation set performance stabilizes, thus balancing model complexity and prediction accuracy. For example, in classification tasks, if the loss function is cross-entropy, a negative gradient is calculated at each step to guide tree construction, and the optimized model can effectively handle imbalanced classes.

[0104] Parameter settings for augmented tree models involve multiple hyperparameters, primarily tuned through cross-validation or grid search. Key parameters include learning rate, number of trees, maximum depth, minimum number of split samples, subsampling ratio, and regularization parameter. When setting these parameters, a coarse-tuning process is first performed by fixing the learning rate and number of trees, followed by fine-tuning of the depth and sampling parameters. Optimization methods can employ random search or Bayesian optimization to efficiently explore the parameter space. For example, when processing a house price prediction dataset, the learning rate can be set to 0.1, the number of trees to 100, and the maximum depth to 3. Cross-validation can then be used to evaluate the root mean square error to determine the optimal combination, thereby improving the model's generalization performance on noisy data.

[0105] In the above embodiment, risk estimation is performed using an enhanced tree model based on the fluctuation spike amount and temperature acoustic deviation rate to obtain the diffusion coefficient of each feeder node; based on the diffusion coefficient of each feeder node, the risk score of each feeder node is determined, thereby improving the accuracy of risk scoring.

[0106] In some embodiments, determining the risk score of each feeder node based on the diffusion coefficient of each feeder node includes: determining the line diffusion coefficient of each edge in the offset fingerprint map based on the diffusion coefficient of each feeder node in the offset fingerprint map; determining multiple node paths based on the offset fingerprint map, starting from each feeder node; determining the risk gradient of each node path based on the line diffusion coefficient of each edge in each node path; and determining the risk score of each feeder node based on the risk gradient of the multiple node paths.

[0107] Specifically, for each edge in the offset fingerprint map, the line diffusion coefficient of the edge is determined based on the diffusion coefficients of the two feeder nodes connected to the edge. For example, the line diffusion coefficient of the edge is obtained by averaging the diffusion coefficients of the two feeder nodes connected to the edge.

[0108] Starting from each feeder node, based on the offset fingerprint map, traverse adjacent nodes along the conductor direction until the source node is reached, obtaining multiple node paths for that feeder node. For each node path, determine the product of the line distance and the line diffusion coefficient corresponding to each edge in the node path to obtain the risk gradient of that edge. Accumulate the risk gradients of multiple edges in the node path to obtain the risk score of that feeder node.

[0109] Here, the risk gradient represents the combined diffusion effect of all edges on the node path of the feeder node diffusion.

[0110] In the above embodiments, multiple node paths are determined based on the offset fingerprint map, starting from each feeder node; and a risk score is determined based on the line diffusion coefficient of each edge in each node path. This simulates the propagation of anomalies along the line, quantifies the potential impact, and improves the accuracy of risk scoring.

[0111] Optionally, after determining the feeder node maintenance queue, the drift duration is predicted based on the diffusion coefficient and the reference propagation speed. For each feeder node, multiple node paths are determined based on the offset fingerprint map, starting from each feeder node. The predicted duration of each edge is obtained by dividing the wire distance of each edge in the node path by the reference propagation speed and then multiplying it by the diffusion coefficient. The predicted duration of each edge is accumulated to obtain the drift duration of the feeder node.

[0112] Drift duration can be used to reflect the time required for a fault to propagate to the feeder node. By guiding emergency repairs through drift duration, emergency repairs can be made more timely.

[0113] In some embodiments, reference Figure 2 The detection method for feeder nodes in the distribution network also includes: S201, determining the first feeder node in the feeder node maintenance queue; S202, obtaining the target multi-source node data of the first feeder node; S203, updating the risk score of each feeder node based on the target multi-source node data and the multi-source node data of the first feeder node, and obtaining the updated risk score of each feeder node; S204, determining the updated feeder node maintenance queue based on the updated risk score of each feeder node.

[0114] Specifically, the first feeder node is the feeder node that is first in the feeder node maintenance queue and also the feeder node with the highest risk score; a sampling command is issued to the first feeder node to collect the target multi-source node data of the first feeder node, that is, to obtain the current multi-source node data of the first feeder node.

[0115] Using the target multi-source node data as the return evidence, the target multi-source node data is time-aligned, and the updated fluctuation spike amount and updated temperature acoustic deviation rate are recalculated based on the target multi-source node data. Based on the fluctuation spike amount and temperature acoustic deviation rate calculated based on the multi-source node data of the first and second feeder nodes, as well as the recalculated updated fluctuation spike amount and updated temperature acoustic deviation rate, the update increment of the risk gradient is calculated; for example, as shown in formula (9).

[0116] Formula (9):

[0117] ;

[0118] in, It updates the gradient increment. It updates the peak value of fluctuations. Indicates the peak value of the fluctuation. This indicates the updated temperature voiceprint deviation rate. This indicates the temperature-to-sound-text deviation rate. This indicates the distance of the line connecting the first and second feeder nodes.

[0119] The updated diffusion coefficient is obtained by correcting the diffusion coefficient of the first feeder node with the updated risk gradient increment; for example, as shown in formula (10).

[0120] Formula (10): ;

[0121] in, It updates the diffusion coefficient. It is the diffusion coefficient. It updates the gradient increment. It is the distance between the source node and the first feeder node.

[0122] Based on the update diffusion coefficient of the first feeder node, the risk scores of all feeder nodes are recalculated to obtain the update risk score of each feeder node. Then, the maintenance queue of the updated feeder nodes is determined based on the update risk score of each feeder node.

[0123] Optionally, the feeder node maintenance queue is compared with the updated feeder node maintenance queue to detect whether the maintenance order of the feeder nodes has changed (detect whether the risk score has changed). If it is determined that the maintenance order has changed, the change is quantified according to the flip index, as shown in formula (11).

[0124] Formula (11): ;

[0125] Here, Flip represents the flip metric. This represents the total number of feeder nodes in the risk queue. Indicates the first Each feeder node is updating the sequence number in the feeder node maintenance queue. Indicates the first The sequence number of each feeder node in the feeder node maintenance queue.

[0126] The magnitude of the sequential change is quantified by flipping the index, so as to intuitively determine the impact of the updated parameters on the propagation path.

[0127] Optionally, the updated feeder node maintenance queue can be sent to the dispatching terminal via a communication interface, allowing the dispatching terminal to receive the update in real time and perform emergency repairs based on the updated feeder node maintenance queue. Real-time updates of the maintenance queue in the distribution network facilitate rapid response by operation and maintenance personnel.

[0128] Optionally, the updated adjacency fingerprint vector and its own difference vector of the first node in the feeder node maintenance queue are stored to provide an update basis for the next round of iterative diffusion, thereby realizing graph diffusion through cyclic execution and real-time capture of anomaly propagation.

[0129] In a specific example, refer to Figure 3 The detection method for feeder nodes in the distribution network is applied to the detection system, which includes a topology database, data acquisition unit, relationship mapping unit, risk analysis unit, correction unit, node sensor cluster, queue update unit, and dispatch terminal; the detection method for feeder nodes includes:

[0130] The data acquisition unit acquires multi-source node data for each feeder node in the distribution network over a preset time period; the topology of the distribution network is acquired through the topology database.

[0131] The relationship mapping unit performs diffusion processing on the multi-source node data of the topology and each feeder node, and determines the adjacent fingerprint vector and its own difference vector corresponding to the diffusion process, thereby generating an offset fingerprint map.

[0132] The risk analysis unit calculates the risk based on the offset fingerprint map to obtain the risk score for each feeder node, and determines the maintenance queue for each feeder node based on the risk score.

[0133] The calibration unit identifies the first feeder node in the feeder node maintenance queue and issues a sampling command to the first feeder node.

[0134] The target multi-source node data of the first feeder node is obtained through the node sensor cluster;

[0135] The correction unit recalculates the updated fluctuation spike and updated temperature acoustic deviation rate based on the multi-source node data of the first feeder node, corrects the diffusion coefficient of the first feeder node, obtains the updated diffusion coefficient, and recalculates the risk score of all feeder nodes based on the updated diffusion coefficient of the first feeder node, thereby obtaining the updated risk score of each feeder node and determining the updated feeder node maintenance queue.

[0136] Send the updated feeder node maintenance queue to the queue update unit, as well as update the fluctuation spike amount and update the temperature acoustic deviation rate;

[0137] The queue update unit sends the update fluctuation spike amount and update temperature acoustic fingerprint deviation rate to the relation mapping unit so that the relation mapping unit can update the offset fingerprint map.

[0138] The queue update unit sends an updated feeder node maintenance queue to the dispatcher so that the dispatcher can carry out emergency repairs based on the updated feeder node maintenance queue.

[0139] The method for detecting feeder nodes in a distribution network provided in this application uses multi-source node data of each feeder node in the distribution network over a preset time period, along with the topology of the distribution network, to perform iterative diffusion processing to obtain the offset characteristics of each feeder node. An offset fingerprint map is generated based on these offset characteristics, quantifying the spread of abnormal situations between feeder nodes. Risk estimation is performed based on the offset fingerprint map to obtain a risk score for each feeder node, resulting in a feeder node maintenance queue. This queue not only reflects high-risk fault sources but also the spread path of faults between feeder nodes. Emergency repairs are performed by referring to the feeder node maintenance queue, improving the success rate of emergency repairs, optimizing the allocation of emergency repair resources, and ensuring the stable operation of the distribution network.

[0140] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0141] Figure 4 This is a schematic diagram of the structure of the detection device for feeder nodes in the distribution network provided in this application, as shown below. Figure 4 As shown, the detection device 40 for feeder nodes in the distribution network provided in this embodiment includes:

[0142] The acquisition module 410 is used to acquire multi-source node data of each feeder node in the distribution network for a preset time period;

[0143] The iterative diffusion module 420 is used to perform iterative diffusion processing based on the multi-source node data of each feeder node and the topology of the distribution network to obtain the offset characteristics of each feeder node.

[0144] The offset fingerprint generation module 430 is used to generate an offset fingerprint based on the offset characteristics of each feeder node and the topology of the distribution network.

[0145] The risk scoring determination module 440 is used to perform risk estimation based on the offset fingerprint map to obtain the risk score of each feeder node.

[0146] The maintenance queue determination module 450 is used to determine the maintenance queue of each feeder node based on the risk score of each feeder node.

[0147] In some embodiments, the iterative diffusion module 420 is configured to: determine the initial multi-source vector of each feeder node based on the multi-source node data of each feeder node; determine the adjacent nodes of each feeder node based on the topology of the distribution network; perform iterative diffusion processing based on the initial multi-source vector of each feeder node and the initial multi-source vector of the adjacent nodes to obtain the diffused multi-source vector of each feeder node; and determine the offset characteristics of each diffusion node based on the initial multi-source vector and the diffused multi-source vector of each feeder node.

[0148] In some embodiments, the iterative diffusion module 420 is configured to determine the adjacent fingerprint vector corresponding to the diffusion process based on the initial multi-source vector and the diffusion multi-source vector of each feeder node; determine its own difference vector based on the initial multi-source vector and the diffusion multi-source vector of each feeder node; and determine the offset feature of each diffusion node based on the adjacent fingerprint vector and its own difference vector of each feeder node.

[0149] In some embodiments, the risk scoring determination module 440 is used to determine the fluctuation spike amount based on the voltage-current difference sequence; determine the temperature-soundprint deviation rate based on the temperature offset sequence and the soundprint offset sequence; perform risk estimation through an enhancement tree model based on the fluctuation spike amount and the temperature-soundprint deviation rate to obtain the diffusion coefficient of each feeder node; and determine the risk score of each feeder node based on the diffusion coefficient of each feeder node.

[0150] In some embodiments, the risk scoring determination module 440 is configured to determine the line diffusion coefficient of each edge in the offset fingerprint map based on the diffusion coefficient of each feeder node in the offset fingerprint map; determine multiple node paths based on the offset fingerprint map, starting from each feeder node; determine the risk gradient of each node path based on the line diffusion coefficient of each edge in each node path; and determine the risk score of each feeder node based on the risk gradient of the multiple node paths.

[0151] In some embodiments, the detection device for feeder nodes in a distribution network further includes: a queue update module, used to determine the first feeder node in the feeder node maintenance queue; acquire target multi-source node data of the first feeder node; update the risk score of each feeder node based on the target multi-source node data and the multi-source node data of the first feeder node to obtain an updated risk score for each feeder node; and determine an updated feeder node maintenance queue based on the updated risk score of each feeder node.

[0152] The detection device for feeder nodes in the distribution network provided in this embodiment can execute the detection method for feeder nodes in the distribution network provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0153] Figure 5 A schematic diagram of the structure of the electronic device provided in this application. Figure 5 As shown, the electronic device 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus.

[0154] In a specific implementation, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to perform the above-described method.

[0155] The specific implementation process of processor 501 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0156] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0157] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0158] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0159] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0160] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0161] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0162] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0163] The division of units is merely a logical functional division; 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 coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0164] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0165] In addition, the functional units in the various embodiments of the present invention 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.

[0166] If a function 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, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0167] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0168] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for detecting feeder nodes in a distribution network, characterized in that, include: Acquire multi-source node data for each feeder node in the distribution network over a preset time period; Based on the multi-source node data of each feeder node and the topology of the distribution network, iterative diffusion processing is performed to obtain the offset characteristics of each feeder node. An offset fingerprint is generated based on the offset characteristics of each feeder node and the topology of the distribution network. Risk assessment is performed based on the offset fingerprint to obtain a risk score for each feeder node; The maintenance queue for each feeder node is determined based on its risk score.

2. The method according to claim 1, characterized in that, The iterative diffusion process, based on the multi-source node data of each feeder node and the topology of the distribution network, yields the offset characteristics of each feeder node, including: Based on the multi-source node data of each feeder node, determine the initial multi-source vector of each feeder node; Based on the topology of the power distribution network, determine the adjacent nodes of each feeder node; Iterative diffusion processing is performed based on the initial multi-source vector of each feeder node and the initial multi-source vector of the adjacent node to obtain the diffused multi-source vector of each feeder node. The offset characteristics of each diffusion node are determined based on the initial multi-source vector of each feeder node and the diffusion multi-source vector.

3. The method according to claim 2, characterized in that, The step of determining the offset characteristics of each diffusion node based on the initial multi-source vector of each feeder node and the diffusion multi-source vector includes: Based on the initial multi-source vector of each feeder node and the diffusion multi-source vector, determine the adjacent fingerprint vector corresponding to the diffusion process; Each feeder node determines its own difference vector based on its initial multi-source vector and the diffusion multi-source vector. The offset features of each diffusion node are determined based on the adjacent fingerprint vector and the self-difference vector of each feeder node.

4. The method according to claim 1, characterized in that, The risk assessment based on the offset fingerprint map, to obtain a risk score for each feeder node, includes: Based on the offset fingerprint, determine the voltage-current differential sequence, temperature offset sequence, and acoustic offset sequence for each feeder node; The fluctuation spike amount is determined based on the voltage-current difference sequence; The temperature-soundprint deviation rate is determined based on the temperature offset sequence and the soundprint offset sequence. Based on the fluctuation spike amount and the temperature acoustic deviation rate, the diffusion coefficient of each feeder node is obtained by risk estimation through the enhancement tree model; The risk score for each feeder node is determined based on the diffusion coefficient of each feeder node.

5. The method according to claim 4, characterized in that, The process of determining the risk score for each feeder node based on its diffusion coefficient includes: Based on the diffusion coefficient of each feeder node in the offset fingerprint, the line diffusion coefficient of each edge in the offset fingerprint is determined. Starting from each feeder node, multiple node paths are determined based on the offset fingerprint map; The risk gradient of each node path is determined based on the path diffusion coefficient of each edge in each node path. Based on the risk gradient of multiple node paths, determine the risk score of each feeder node.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: The first feeder node is determined in the feeder node maintenance queue; Obtain the target multi-source node data of the first feeder node; Based on the target multi-source node data and the multi-source node data of the first feeder node, the risk score of each feeder node is updated to obtain the updated risk score of each feeder node. The maintenance queue for each feeder node is determined based on its update risk score.

7. A detection device for feeder nodes in a power distribution network, characterized in that, include: The acquisition module is used to acquire multi-source node data for each feeder node in the distribution network for a preset time period; The iterative diffusion module is used to perform iterative diffusion processing based on the multi-source node data of each feeder node and the topology of the distribution network to obtain the offset characteristics of each feeder node. The offset fingerprint generation module is used to generate an offset fingerprint based on the offset characteristics of each feeder node and the topology of the distribution network. The risk scoring determination module is used to perform risk estimation based on the offset fingerprint map to obtain the risk score for each feeder node. The maintenance queue determination module is used to determine the maintenance queue of each feeder node based on the risk score of each feeder node.

8. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes computer execution instructions, which, when executed by a processor, implement the method as described in any one of claims 1 to 6.

Images