Method and device for processing data transmission fault of distribution network terminal and electronic equipment

By acquiring and analyzing transmission data in the distribution network terminal, extracting key features and determining the fault location using causal network and federated learning, the problem of difficult to identify complex fault patterns in the existing technology is solved, and the accuracy of distribution network failures is achieved Diagnosis and repair improve diagnostic capabilities and data security.

CN119052055APending Publication Date: 2024-11-29GUANGDONG POWER GRID CO LTD +1
View PDF 0 Cites 5 Cited by

Patent Information

Application Number
CN202411044276.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

In the face of complex or unknown fault modes, existing distribution network automation systems are difficult to accurately identify the cause of the fault, resulting in low diagnostic accuracy and long response time, and lack of data privacy and security.

Method used

By obtaining multiple transmission data from distribution network terminals, log analysis and abnormal detection, extracting key features, determining dynamic thresholds, using causal network analysis and federated learning to determine the location of the fault, and generating fault repair strategies to achieve accurate location and repair of faults .

Benefits of technology

Significantly improve the ability to diagnose complex and unknown fault patterns, accurately distinguish the causal relationships and simple correlations between fault phenomena, ensure data privacy and share learning results, accurately locate the cause of failures, and solve existing technologies It is difficult to identify the cause of data transmission failure in distribution network.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119052055A_ABST
    Figure CN119052055A_ABST
Patent Text Reader

Abstract

The invention provides a processing method and device for a data transmission fault of a distribution network terminal and electronic equipment. The method comprises the following steps: acquiring various transmission data of a distribution network terminal, and performing log analysis on the transmission data to obtain a first anomaly detection result; extracting key features of the plurality of transmission data, determining a dynamic threshold according to the key features, and performing anomaly detection to obtain a second anomaly detection result; analyzing the first anomaly detection result and the second anomaly detection result by using a causal network, determining an initial position of a data transmission fault of the distribution network terminal, implementing federated learning among a plurality of nodes in the causal network, and updating the data transmission fault of the distribution network terminal by adjusting weight distribution and parameter updating strategies in the federated learning. And further determining the final position of the data transmission fault of the distribution network terminal, and generating a fault repair strategy according to the final position to repair the data transmission fault of the distribution network terminal. According to the invention, the problem that the data transmission fault position of the power distribution network is difficult to identify in the prior art is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of distribution network terminal fault processing, and in particular, to a distribution network terminal data transmission fault processing method, a distribution network terminal data transmission fault processing device, a computer-readable storage medium and an electronic device. Background Art

[0002] With the rapid development of power system automation technology, distribution network automation has become one of the key technologies to improve grid reliability, optimize resource allocation and enhance user service quality. Especially in large-scale power grids such as the Southern Power Grid, the widespread deployment of distribution automation terminals (such as intelligent terminals, remote control terminals, etc.) greatly relies on stable and efficient data transmission systems. However, existing distribution automation systems face a series of technical challenges and limitations, especially in data transmission fault diagnosis.

[0003] At present, most distribution automation systems use standardized communication mechanisms based on protocols such as the 101 communication protocol. These systems often use certain traditional fault detection and diagnosis methods, such as simple threshold determination and periodic status checks. Although these methods perform well in dealing with common and predefined fault types, they have low diagnostic accuracy and long response time when facing complex or unknown fault modes, making it difficult to meet the requirements of efficiency and real-time performance. For example, in the case of communication disconnection or instability, existing technologies often have difficulty in accurately identifying the root cause of the fault, resulting in the inability to carry out targeted repair work, affecting the stable operation of the entire power grid.

[0004] In addition, existing technologies also have deficiencies in data privacy and security. During the fault diagnosis process, a large amount of terminal data needs to be collected and transmitted, which not only increases the risk of data leakage, but also increases the burden of data processing. Therefore, a new technical solution is needed that can effectively solve the above problems and improve the efficiency and accuracy of fault diagnosis. Summary of the invention

[0005] The main purpose of the present application is to provide a method for processing data transmission failure of a distribution network terminal, a device for processing data transmission failure of a distribution network terminal, a computer-readable storage medium and an electronic device, so as to at least solve the problem in the prior art that it is difficult to identify the cause of data transmission failure in the distribution network.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for processing data transmission failure of a distribution network terminal is provided, including: obtaining multiple transmission data of the distribution network terminal, performing log analysis on the transmission data, and obtaining a first abnormality detection result; extracting key features of the multiple transmission data, determining a dynamic threshold based on the key features, and performing abnormality detection to obtain a second abnormality detection result, wherein the key feature is a feature that characterizes the data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure; using a causal network to analyze the first abnormality detection result and the second abnormality detection result, determine the initial position of the data transmission failure of the distribution network terminal, implement federated learning between multiple nodes in the causal network, further determine the final position of the data transmission failure of the distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning, and generate a fault repair strategy based on the final position to repair the data transmission failure of the distribution network terminal.

[0007] Optionally, using a causal network to analyze the first anomaly detection result and the second anomaly detection result to determine the initial location of the data transmission failure of the distribution network terminal, including: obtaining time series data of the sensor of the distribution network terminal and synchronizing the timestamp, defining each of the sensors as a node, and determining the connection relationship between multiple nodes according to the physical connection relationship of the power grid and the liquidity of historical data; using the transfer entropy method Calculating the causal relationship between the plurality of nodes to obtain a transfer entropy value, and constructing a directed graph according to the transfer entropy value, wherein the edge of the directed graph is determined by the size of the transfer entropy value and represents the information flow or influence from one node to another node, represents the state of the node at time t, represents the state of the edge at time t, represents the state of the edge at time t+1, TE i→j Representing the transfer entropy value; determining the node where the abnormality first occurs as the initial starting point of the data transmission failure, and using a graph search algorithm in the directed graph to determine multiple fault propagation paths starting from the initial starting point to obtain multiple initial positions.

[0008] Optionally, extracting a plurality of key features of the transmission data includes: using an empirical mode decomposition method EMD(X(t))={IMF 1 ,IMF 2 ,......,IMF n ,R} decompose the transmission data to obtain multiple intrinsic modes and a residual term, where IMF i represents the ith intrinsic mode, R represents the residual term, and X(t) represents the input signal; the signal is reconstructed according to the intrinsic mode To filter high frequency noise and extract the key features of the transmission data, where k represents the number of the natural modes, X rec (t) represents the reconstructed signal.

[0009] Optionally, determining a dynamic threshold according to the key feature and performing anomaly detection to obtain a second anomaly detection result includes: calculating the power spectrum of the reconstructed signal by a fast Fourier transform method to obtain the spectral entropy of the reconstructed signal as follows: Where p(f) is the normalized power spectral density, F represents the total number of frequency components, SE represents the spectral entropy, and the reconstructed signal is a signal obtained by performing empirical mode decomposition and signal reconstruction on the key feature; the moving average threshold and moving standard deviation threshold of the key feature are calculated according to the spectral entropy as follows: Wherein, α represents the first smoothing factor, β represents the second smoothing factor, the first smoothing factor and the second smoothing factor are used for the weight of historical data on the current estimate, t represents time, μ SE represents the moving average threshold, σ SE Indicates the moving standard deviation threshold; when the current spectral entropy meets the condition In the case of, determining that the transmission data is abnormal, wherein SE t Represents the current spectral entropy corresponding to time t.

[0010] Optionally, federated learning is implemented among multiple nodes in the causal network, and the final location of the data transmission failure of the distribution network terminal is further determined by adjusting the weight distribution and parameter update strategy in the federated learning, including: obtaining the historical data quality and reliability of the node, and calculating the score Score according to the historical data quality and the reliability i =a×Completeness i +b×Synchrony i , the node with the highest calculation score is used for training the federated learning model, where Score i represents the score of node i, a represents the quality coefficient of historical data, b represents the reliability coefficient, and completeness i represents the quality of the historical data of the node i, Synchrony i Represents the reliability of the node i; adjusts the learning rate η of the federated learning model new =η old ×exp(-γ×ΔL), where η old represents the learning rate before adjustment, η newrepresents the adjusted learning rate, γ represents the adjustment factor of the learning rate, and ΔL represents the loss reduction of two consecutive iterations; the model parameters of each node are aggregated using an adaptive weighted average method To update the federated learning model, where w i represents the data quality of the node i, θ i represents the model parameters of the node i, N represents the total number of the nodes, θ global represents the model parameters after aggregation.

[0011] Optionally, a fault repair strategy is generated according to the final position, including: acquiring real-time current data and real-time voltage data, applying a time series anomaly detection algorithm to identify potential fault signals in real time, and using a machine learning classification model to automatically classify the fault types of the potential fault signals; constructing an influence propagation model based on graph theory to simulate the influence diffusion process starting from the fault point, determining a priority algorithm based on fault impact and repair difficulty according to the influence diffusion process, and calculating the fault repair strategy according to the priority algorithm.

[0012] Optionally, after obtaining multiple transmission data of the distribution network terminal, the method also includes: low-pass filtering the transmission data through a fourth-order Butterworth low-pass filter to obtain low-pass filtered data to remove high-frequency noise greater than the power grid frequency; and smoothing the low-pass filtered data using a sliding window to obtain the pre-processed transmission data.

[0013] According to another aspect of the present application, a device for processing data transmission failure of a distribution network terminal is provided, comprising: an analysis unit, used to obtain multiple transmission data of the distribution network terminal, perform log analysis on the transmission data, and obtain a first abnormality detection result; a detection unit, used to extract key features of the multiple transmission data, determine a dynamic threshold based on the key features, and perform abnormality detection to obtain a second abnormality detection result, wherein the key feature is a feature that characterizes the data transmission failure, and the dynamic threshold is used to determine whether there is a failure in the transmission data; a generation unit, used to use a causal network to analyze the first abnormality detection result and the second abnormality detection result, determine the initial position of the data transmission failure of the distribution network terminal, implement federated learning between multiple nodes in the causal network, further determine the final position of the data transmission failure of the distribution network terminal by adjusting the weight allocation and parameter update strategy in the federated learning, and generate a fault repair strategy based on the final position to repair the data transmission failure of the distribution network terminal.

[0014] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the methods for processing data transmission failures of distribution network terminals.

[0015] According to another aspect of the present application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing any one of the processing methods for data transmission failures of distribution network terminals.

[0016] Apply the technical solution of the present application, obtain multiple transmission data of the distribution network terminal, perform log analysis on the transmission data, and obtain a first abnormal detection result; extract key features of multiple transmission data, determine dynamic thresholds according to the key features, and perform abnormal detection to obtain a second abnormal detection result, wherein the key features are features that characterize data transmission failures, and the dynamic thresholds are used to determine whether there is a failure in the transmission data; use the causal network to analyze the first abnormal detection result and the second abnormal detection result, determine the initial position of the data transmission failure of the distribution network terminal, implement federated learning between multiple nodes in the causal network, and further determine the final position of the data transmission failure of the distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning, and generate a fault repair strategy according to the final position to repair the data transmission failure of the distribution network terminal. Compared with the prior art, which is difficult to identify the cause of the data transmission failure of the distribution network, this method significantly improves the diagnostic ability of complex and unknown fault modes by automatically learning and extracting key features, can accurately distinguish the causal relationship and simple correlation between the fault phenomena, and uses the federated learning framework to enable multiple distribution terminals to share learning results under the premise of ensuring data privacy, thereby accurately locating the cause of the failure. Therefore, it can solve the problem that it is difficult to identify the cause of the data transmission failure of the distribution network in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 A hardware structure block diagram of a mobile terminal for executing a method for processing a data transmission failure of a distribution network terminal provided in an embodiment of the present application is shown;

[0019] Figure 2 A flow chart of a method for processing a data transmission failure of a distribution network terminal provided in an embodiment of the present application is shown;

[0020] Figure 3 A flowchart of a specific method for processing a data transmission failure of a distribution network terminal provided by an embodiment of the present application is shown;

[0021] Figure 4 A structural block diagram of a device for processing data transmission failure of a distribution network terminal provided in an embodiment of the present application is shown.

[0022] The above drawings include the following reference numerals:

[0023] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0024] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0025] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] As introduced in the background technology, it is difficult to identify the location of distribution network data transmission failure in the prior art. In order to solve the problem of difficulty in identifying the cause of distribution network data transmission failure, the embodiments of the present application provide a method for processing distribution network terminal data transmission failure, a processing device for distribution network terminal data transmission failure, a computer-readable storage medium and an electronic device.

[0028] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0029] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal of a method for processing a data transmission failure of a distribution network terminal according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0030] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for processing data transmission failure of the distribution network terminal in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the mobile terminal via a network. Examples of the above networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof. The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0031] In this embodiment, a method for processing data transmission failure of a distribution network terminal running on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0032] Figure 2 1 is a flow chart of a method for processing a data transmission failure of a distribution network terminal according to an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:

[0033] Step S201, obtaining various transmission data of the distribution network terminal, performing log analysis on the transmission data, and obtaining a first abnormality detection result;

[0034] Specifically, this step serves as a basis for efficiently collecting and integrating multi-source data (including current, voltage, temperature, and communication logs) from various distribution terminals, ensuring the data integrity and quality for subsequent analysis. This data integration technology automatically processes and fuses different data sources, provides accurate input for fault detection and diagnosis, and solves the problem of data dispersion and inconsistency in the background technology. In the specific implementation, the timestamps of all sensors will be synchronized through the Network Time Protocol (NTP) to ensure the time consistency of data collection. Set the standard time t_standard and record it in Unix time. Using synchronous sampling technology, the sampling frequency is set to f_s = 1kHz, which is sufficient to capture changes caused by grid frequency changes and transient faults. Current I(t) and voltage V(t) data will be continuously recorded by smart meters with built-in AD converters. Specifically, interpolation is used to align all data streams. For data collected at different frequencies, such as temperature, which is usually once a minute, and current and voltage, which are several times per second, linear interpolation will be used to resample the temperature data to align it with the current and voltage data timestamps. For any time point t', the interpolation of current data is expressed as: Among them, t and t+Δt are known data points, and t' is the time point that needs to be interpolated. In the specific implementation, the above preprocessing is signal filtering, including: low-pass filtering is represented by implementing a fourth-order Butterworth low-pass filter for current and voltage signals, with a cutoff frequency set to 50Hz to remove high-frequency noise above the grid frequency. The filter formula is: Among them, ω c =2π×50Hz, n=4. The smoothing process is to apply sliding window average filtering to the temperature data. The window size is set to 5 minutes, corresponding to 5 data points. The smoothed temperature is: Then, data quality control and anomaly detection are performed. Specifically, outlier detection is performed on all data streams. The moving average μ and standard deviation σ of each data type are calculated, and data points outside the range of μ±3σ are dynamically identified and marked as abnormal. Error and warning messages in communication logs are automatically analyzed, combined with abnormal fluctuations in current, voltage, and temperature data, to identify potential equipment failures or data transmission problems.

[0035] Step S202, extracting key features of the plurality of transmission data, determining a dynamic threshold value according to the key features, and performing anomaly detection to obtain a second anomaly detection result, wherein the key features are features characterizing a data transmission failure, and the dynamic threshold value is used to determine whether the transmission data has a failure;

[0036] Specifically, in the specific implementation, this step can automatically learn and extract key features that are helpful for fault diagnosis from unlabeled data. This learning mode does not rely on pre-defined labels, reduces the need for manual input, and improves the ability to handle unknown fault modes. Through methods such as contrast learning, the system can automatically mark potential faults, significantly improve the sensitivity and accuracy of fault detection, and solve the shortcomings of traditional methods in identifying complex faults. Among them, the empirical mode decomposition (EMD) method is applied to decompose the time series data I(t) and V(t) into a series of intrinsic mode functions (IMFs) and a residual term. The method extracts the oscillation mode of the signal step by step through the interpolation calculation of local extreme values ​​to adapt to the characteristics of nonlinear and non-stationary signals. In the specific implementation, the present invention also provides a comprehensive data management method for data marking and recording. Specifically, for each detected anomaly, detailed time information is recorded, including the exact start time and end time of the anomaly, and the duration. At the same time, all key information related to the anomaly is recorded, including spectral entropy values, selected IMFs, environmental conditions (such as temperature, humidity) when the anomaly occurs, and equipment status. Finally, a real-time updated data dashboard is provided to display current and historical abnormal events, including its spectrum entropy change trend graph, affected power lines and equipment. Developing APIs allows the system to be integrated with other fault management and monitoring systems to achieve instant data sharing and multi-system collaboration.

[0037] Step S203, use a causal network to analyze the above-mentioned first anomaly detection result and the above-mentioned second anomaly detection result, determine the initial position of the data transmission failure of the above-mentioned distribution network terminal, implement federated learning between multiple nodes in the above-mentioned causal network, and further determine the final position of the data transmission failure of the above-mentioned distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning, and generate a fault repair strategy based on the above-mentioned final position to repair the data transmission failure of the above-mentioned distribution network terminal.

[0038] Specifically, in the specific implementation, this step applies causal inference technology to analyze the results of anomaly detection and determine the specific cause and precise location of the fault. This graph-based causal inference algorithm can accurately identify the source of the fault and the propagation path in the complex system of the power grid. It solves the problem that the traditional method only relies on correlation analysis and cannot accurately locate and diagnose the cause of the fault, thereby significantly improving the efficiency and effectiveness of fault handling.

[0039] Through this embodiment, a variety of transmission data of the distribution network terminal is obtained, and the transmission data is log analyzed to obtain a first abnormal detection result; the key features of the various transmission data are extracted, and the dynamic threshold is determined according to the key features, and an abnormal detection is performed to obtain a second abnormal detection result, wherein the key feature is a feature that characterizes the data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure; the first abnormal detection result and the second abnormal detection result are analyzed using a causal network to determine the initial position of the data transmission failure of the distribution network terminal, and federated learning is implemented between multiple nodes in the causal network. By adjusting the weight distribution and parameter update strategy in the federated learning, the final position of the data transmission failure of the distribution network terminal is further determined, and a fault repair strategy is generated according to the final position to repair the data transmission failure of the distribution network terminal. Compared with the prior art, which is difficult to identify the cause of the distribution network data transmission failure, this method significantly improves the diagnostic ability of complex and unknown fault modes by automatically learning and extracting key features, and can accurately distinguish the causal relationship and simple correlation between the fault phenomena. By using the federated learning framework, multiple distribution terminals can share learning results under the premise of ensuring data privacy, thereby accurately locating the cause of the failure. Therefore, the problem that it is difficult to identify the cause of the distribution network data transmission failure in the prior art can be solved.

[0040] In the specific implementation process, the above step S203 uses a causal network to analyze the above first abnormality detection result and the above second abnormality detection result to determine the initial location of the data transmission failure of the above distribution network terminal, which can be achieved by the following steps: Step S2031: Obtain the time series data of the sensor of the above distribution network terminal and synchronize the timestamp, define each of the above sensors as a node, and determine the connection relationship between multiple nodes according to the physical connection relationship of the power grid and the liquidity of historical data; Step S2032: Use the transfer entropy method Calculate the causal relationship between the multiple nodes to obtain the transfer entropy value, and construct a directed graph based on the transfer entropy value, wherein the edge of the directed graph is determined by the size of the transfer entropy value and represents the information flow or influence from one node to another. represents the state of the node at time t, represents the state of the edge at time t, represents the state of the above edge at time t+1, TE i→jrepresents the above transfer entropy value; Step S2033: determine the node where the abnormality first occurs as the initial starting point of the data transmission fault, and use a graph search algorithm in the above directed graph to determine multiple fault propagation paths from the above initial starting point to obtain multiple above initial positions. The method determines the initial position of the fault through the above steps, so that the position of the fault can be further located based on the initial position.

[0041] Specifically, this step includes: collecting multivariate time series data of key nodes, such as current, voltage and environmental parameters (such as temperature), and ensuring the temporal consistency of all data. Specifically, multifunctional sensors are installed at key nodes of the distribution network, which can measure parameters such as current, voltage and ambient temperature, and can send data in real time. The global positioning system (GPS) or network time protocol (NTP) is used to ensure the synchronization of timestamps of all sensors so as to accurately capture and analyze the temporal relationship in the data. A set of safe and reliable data transmission systems is established to transmit the data collected by the sensors to a central database in real time. Encrypted transmission is used to ensure the security of data during transmission. A dynamic causal network is constructed based on the collected data to understand the dependencies and causal relationships between different nodes. For the present invention, each sensor or power grid component is defined as a node in the graph, and the connection relationship between nodes is defined according to the physical connection of the power grid and the mobility of historical data. The causal relationship between nodes is calculated using the transfer entropy method. The transfer drop not only measures the flow of information between the two systems, but also reflects the dynamic dependency between them. Identify the starting point of the fault and infer the propagation path of the fault in the network. Specifically, based on the anomaly detection results (data analysis from the previous step), identify the node where the anomaly first occurs as the possible starting point of the fault. Using the constructed causal graph, a graph search algorithm (such as depth-first search DFS) is used to trace the possible fault propagation path starting from the fault starting point. The path selection is based on the size of the transfer entropy value. A high transfer entropy value indicates a large information flow or influence, and a higher possibility of fault propagation. Generate a detailed fault report based on the analysis results, and use the discovered information to optimize the system's monitoring and diagnosis process. Specifically, record each identified fault in detail, including the fault type, affected nodes, time of fault occurrence, and inferred fault cause and consequences. According to the results of fault handling and subsequent operation and maintenance feedback, adjust the data acquisition parameters and causal analysis model to improve the accuracy of future fault prediction and location. Implement continuous system performance monitoring to evaluate the effectiveness of optimization measures and further adjust strategies.

[0042] In some optional implementations, the above step S202 extracts the key features of the above transmission data, which can be achieved by the following steps: Step S2021: Adopting the empirical mode decomposition method EMD(X(t))={IMF 1 ,IMF 2,......,IMF n ,R} decompose the above transmission data to obtain multiple intrinsic modes and a residual term, among which, IMF i represents the i-th intrinsic mode, R represents the residual term, and X(t) represents the input signal; Step S2022: reconstruct the signal according to the intrinsic mode To filter high frequency noise and extract the above key features of the above transmission data, where k represents the number of the above natural modes, X rec (t) represents the reconstructed signal. This method extracts key features through the above steps, so that the fault position can be located through the key features.

[0043] In the specific implementation process, this step can automatically learn and extract key features that are helpful for fault diagnosis from unlabeled data. This learning model does not rely on pre-defined labels, reduces the need for manual input, and improves the ability to handle unknown fault modes. Through methods such as contrast learning, the system can automatically mark potential faults, significantly improve the sensitivity and accuracy of fault detection, and solve the shortcomings of traditional methods in identifying complex faults. Among them, the empirical mode decomposition (EMD) method is used to decompose the time series data I(t) and V(t) into a series of intrinsic mode functions (IMFs) and a residual term. This method extracts the oscillation mode of the signal step by step through the interpolation calculation of local extreme values ​​to adapt to the characteristics of nonlinear and non-stationary signals.

[0044] In some other optional implementations, the above step S202 determines the dynamic threshold according to the above key features, and performs anomaly detection to obtain the second anomaly detection result, which can be achieved by the following steps: Step S2023: Calculate the power spectrum of the reconstructed signal by the fast Fourier transform method to obtain the spectral entropy of the reconstructed signal as follows: Where p(f) is the normalized power spectral density, F represents the total number of frequency components, SE represents the above spectral entropy, and the above reconstructed signal is a signal obtained by performing empirical mode decomposition and signal reconstruction on the above key features; Step S2024: Calculate the moving average threshold and moving standard deviation threshold of the above key features according to the above spectral entropy: Among them, α represents the first smoothing factor, β represents the second smoothing factor, the first smoothing factor and the second smoothing factor are used for the weight of historical data on the current estimate, t represents time, μ SE represents the above moving average threshold, σ SE represents the above moving standard deviation threshold; Step S2025: when the current spectral entropy meets the condition In the case of, it is determined that the above transmission data is abnormal, among which, SE tThe above current spectral entropy corresponding to time t is represented by the method. The method detects through the dynamic threshold of the above steps, so that anomaly detection can be performed to accurately detect the abnormal position.

[0045] The specific implementation process also includes automatically adjusting the values ​​of α and β to optimize the sensitivity and accuracy of the thresholds based on the accuracy and feedback of past anomaly detection (such as actual fault confirmation). After diagnosis, the weight of historical data is adjusted based on the feedback of actual fault diagnosis results to improve the accuracy of future predictions.

[0046] In some optional implementations, the above step S203 implements federated learning among multiple nodes in the above causal network, and further determines the final location of the data transmission failure of the above distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning. This can be achieved by the following steps: Step S2034: Obtain the historical data quality and reliability of the above node, and calculate the score Score based on the above historical data quality and the above reliability i =a×Completeness i +b×Synchrony i , the node with the highest calculation score is used for training the federated learning model, where Score i represents the score of node i, a represents the quality coefficient of historical data, b represents the reliability coefficient, and completeness i Indicates the quality of the above historical data of the above node i, Synchrony i Represents the reliability of the node i; Step S2035: Adjust the learning rate η of the federated learning model new =η old ×exp(-γ×ΔL), where η old represents the above learning rate before adjustment, η new represents the adjusted learning rate, γ represents the adjustment factor of the learning rate, and ΔL represents the loss reduction of two consecutive iterations; Step S2036: Adopting the adaptive weighted average method to aggregate the model parameters of each node To update the above federated learning model, where w i represents the data quality of the above node i, θ i represents the model parameters of the above node i, N represents the total number of the above nodes, θ global Represents the above model parameters after aggregation. The method is adjusted and updated through the above steps, so that the fault location can be determined more accurately.

[0047] In the specific implementation process, this step specifically includes: adaptive initialization and node selection. Specifically, the node is scored according to the quality and reliability of its historical data, and nodes with high scores are selected to participate in model training. The following scoring formula is used to consider the data integrity and time synchronization of each node. The model parameters are initialized on the central server, and the initial parameter allocation is adjusted according to the node's score to ensure that high-quality nodes have a greater impact on the model. Each node trains the model based on local data and optimizes the learning process through an adaptive algorithm. Specifically, each node uses its local data for model training and optimizes the following loss function: Among them, D local is the local dataset of the node, loss is the loss function, and λ is the regularization parameter. A dynamic adjustment mechanism is introduced to adjust the learning rate according to the performance of the model on the local dataset: η new =η old ×exp(-γ×ΔL), where η old and η new is the learning rate before and after adjustment, γ is the learning rate adjustment factor, ΔL is the loss reduction between two consecutive iterations, and the model parameters of each node are aggregated on the central server, and the global model is updated. Specifically, the model parameters of each node are aggregated using the adaptive weighted average method: Among them, w i is a weight dynamically calculated based on the data quality and model performance of node i, θ i is the model parameter of node i. The aggregated global model parameters are sent back to each node for the next round of training. The weight w of the node based on the performance of the global model i Make adjustments to optimize subsequent aggregation effects. Monitor the performance of the federated learning process and optimize the learning strategy and parameters based on the monitoring results. Specifically, regularly evaluate the performance of the global model on independent test sets, including the accuracy and recall of fault diagnosis. Analyze the performance of the model on different types of fault diagnosis tasks and identify optimization space. Further, adjust parameters such as α, β, and γ based on the performance evaluation results to improve learning efficiency and model accuracy. Update data collection and preprocessing strategies to improve data quality and further strengthen the foundation of model learning.

[0048] In some optional implementations, the above step S203 generates a fault repair strategy based on the above final position, which can be implemented by the following steps: obtaining real-time current data and real-time voltage data, applying a time series anomaly detection algorithm to identify potential fault signals in real time, and using a machine learning classification model to automatically classify the fault types of the above potential fault signals; constructing an influence propagation model based on graph theory to simulate the influence diffusion process starting from the fault point, determining a priority algorithm based on the fault impact and repair difficulty according to the above influence diffusion process, and calculating the above fault repair strategy according to the above priority algorithm. The method generates a fault repair strategy through the above steps, so that the fault can be automatically repaired after the fault is located.

[0049] Specifically, this step includes: 1. Real-time identification of faults in the distribution network, and intelligent classification based on fault characteristics, so as to take targeted response measures. Among them, fault signal identification uses the current and voltage data monitored in real time, and applies the time series anomaly detection algorithm to identify potential fault signals in real time. The fault detection formula is based on the physical laws of electrical systems, such as Ohm's law and Kirchhoff's current law, which are used to calculate node current and voltage deviations: Among them I expected and V expected is the expected value based on network load prediction, I measured and V measuredis the measured value. Intelligent classification of fault types is to automatically classify fault types using machine learning classification models, such as short circuit, open circuit, equipment failure, etc. Model training uses historical fault data and corresponding response effects to enhance classification accuracy. Second, evaluate the impact of faults on distribution network operations, reasonably allocate response priorities, and ensure power supply safety in key areas and key users. Specifically, develop an impact propagation model based on graph theory, which can simulate the impact diffusion process starting from the fault point; wherein, the impact propagation model formula may include: S(t+1)=S(t)·A+B·E(t), where S(t) is the system state vector at time t, A is the adjacency matrix of the network topology, B is the impact diffusion coefficient matrix, and E(t) is the fault impact vector. Further, design a priority algorithm based on fault impact and repair difficulty, which considers both the breadth and depth of fault impact. Priority calculation may include: P=α·Impact+β·Difficulty, where α and β are weight parameters, Impact is the fault impact score, and Difficulty is the repair difficulty score. 3. Automatic resource allocation and fault repair task dispatching, optimize the utilization of human and material resources, and shorten the troubleshooting and recovery time. Specifically, the optimization algorithm is used to automatically determine the best resource allocation plan, including the scheduling of the repair team and the allocation of necessary equipment. The resource optimization model may use linear programming or integer programming techniques to minimize the total cost or total response time. For the present invention, the automatic dispatching of repair tasks is to automatically dispatch repair tasks to the most suitable maintenance team in combination with the fault type and the geographic information system (GIS). The task dispatch algorithm takes into account the team's skills, availability, and distance from the fault point to optimize the response time and repair efficiency.

[0050] In the specific implementation process, an adaptive learning mechanism is introduced to build a real-time adaptive fault diagnosis and learning framework, which automatically adjusts the fault detection and repair strategy according to multi-source data and external environmental changes. In the specific implementation, the system automatically adjusts the fault detection and response strategy according to performance feedback and external environmental changes through an adaptive learning mechanism. This ensures that the system can continuously adapt to new fault modes and environmental changes, solving the problem that traditional systems cannot optimize and update themselves.

[0051] Choose a suitable online learning algorithm as the basis, such as online stochastic gradient descent (SGD) or online recursive least squares estimation (RLS). Design an adaptive learning rate adjustment mechanism to dynamically adjust the learning rate through online performance feedback: Among them, η t is the learning rate at time t, γ is the adjustment factor, is the gradient of the loss function ξ with respect to the parameter θ. Every time new data is received, the model parameters are updated immediately using these data: Among them, xt and t Represent the input data and observed output at time t, respectively, and f is the fault detection model. Furthermore, the impact of the fault on the distribution network is evaluated in real time, and resource allocation is optimized to minimize the negative impact of the fault.

[0052] Specifically, the graph theory model is used to evaluate the possible paths and impact range of fault spread in real time, and the impact degree is calculated in combination with network flow theory: Among them, c n is the importance coefficient of node n, Flow disrupted (n,t) is the amount of traffic interruption due to the fault at time t.

[0053] Then, according to the real-time evaluation results and pre-defined priority rules, technicians and maintenance equipment are dynamically allocated: Allocate(t) = argmin r∈R {Cost(r)+λ·TimeDelay(r,t)}, where R is the set of available resources, Cost(r) and TimeDelay(r,t) represent the cost and possible time delay of using resource r, respectively, and λ is a trade-off factor.

[0054] Furthermore, the system performance is continuously monitored, and the learning strategy and system configuration are adjusted based on performance feedback to ensure that the system is continuously optimized and adapts to grid changes. Among them, a comprehensive monitoring system is implemented to regularly collect data on fault detection accuracy, response time, and user satisfaction. The monitoring data is analyzed using a data-driven approach to identify performance bottlenecks and optimization opportunities. Based on the monitoring results, the model parameters, learning strategies, and resource allocation schemes are automatically adjusted to continuously improve the system efficiency and accuracy.

[0055] In some optional implementations, after obtaining multiple transmission data of the distribution network terminal, the method further includes the following steps: low-pass filtering the transmission data through a fourth-order Butterworth low-pass filter to obtain low-pass filtered data to remove high-frequency noise greater than the power grid frequency; and smoothing the low-pass filtered data using a sliding window to obtain the pre-processed transmission data. The method performs data preprocessing through the above steps, so that the transmission data can be processed into a unified format that is easy to process, so as to facilitate subsequent data processing.

[0056] Specifically, the above preprocessing is signal filtering, including: low-pass filtering is represented by implementing a fourth-order Butterworth low-pass filter for the current and voltage signals, with a cutoff frequency set to 50 Hz to remove high-frequency noise above the grid frequency. The filter formula is: Among them, ω c=2π×50Hz, n=4. The smoothing process is to apply sliding window average filtering to the temperature data. The window size is set to 5 minutes, corresponding to 5 data points. The smoothed temperature is: Then perform data quality control and anomaly detection.

[0057] Specifically, outlier detection is implemented for all data streams. The moving average μ and standard deviation σ of each data type are calculated, and data points outside the range of μ+3σ are dynamically identified and marked as abnormal. Error and warning messages in communication logs are automatically analyzed, combined with abnormal fluctuations in current, voltage, and temperature data, to identify potential equipment failures or data transmission issues.

[0058] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the method for processing data transmission failure of a distribution network terminal of the present application will be described in detail below in combination with specific embodiments.

[0059] This embodiment relates to a method for processing a data transmission failure of a distribution network terminal. Figure 3 As shown, the following steps are included:

[0060] Step S1: Collect multi-source data from distribution network terminals, pre-process the data, and perform outlier detection and communication log analysis on the data;

[0061] Step S2: Use self-supervised learning to extract key features from unlabeled data and mark potential faults. At the same time, design dynamic thresholds based on key features for anomaly detection and perform adaptive adjustments.

[0062] Step S3: using a causal network based on a dynamic causal inference technique to analyze the results of anomaly detection, and determining the cause and location of the data transmission failure at the distribution network terminal;

[0063] Step S4: Implementing federated learning among multiple nodes in the causal network, and adjusting the weight distribution and parameter update strategy in the federated learning, to further determine the location where the network terminal data transmission failure occurs;

[0064] Step S5: automatically formulate a targeted repair strategy according to the location where the network terminal data transmission failure occurs;

[0065] Step S6: Introduce an adaptive learning mechanism, build a real-time adaptive fault diagnosis and learning framework, and automatically adjust the fault detection and repair strategy according to multi-source data and external environment changes.

[0066] The embodiment of the present application also provides a processing device for a distribution network terminal data transmission failure. It should be noted that the processing device for a distribution network terminal data transmission failure in the embodiment of the present application can be used to execute the processing method for a distribution network terminal data transmission failure provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and those that have been described will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0067] The following introduces a device for processing data transmission failure of a distribution network terminal provided in an embodiment of the present application.

[0068] Figure 4 Schematic diagram of a device for processing a data transmission failure of a distribution network terminal according to an embodiment of the present application. Figure 4 As shown, the device comprises:

[0069] The analysis unit 10 is used to obtain various transmission data of the distribution network terminal, perform log analysis on the transmission data, and obtain a first abnormality detection result;

[0070] Specifically, this step serves as a basis for efficiently collecting and integrating multi-source data (including current, voltage, temperature, and communication logs) from various distribution terminals, ensuring the data integrity and quality for subsequent analysis. This data integration technology automatically processes and fuses different data sources, provides accurate input for fault detection and diagnosis, and solves the problem of data dispersion and inconsistency in the background technology. In the specific implementation, the timestamps of all sensors will be synchronized through the Network Time Protocol (NTP) to ensure the time consistency of data collection. Set the standard time t_standard and record it in Unix time. Using synchronous sampling technology, the sampling frequency is set to f_s = 1kHz, which is sufficient to capture changes caused by grid frequency changes and transient faults. Current I(t) and voltage V(t) data will be continuously recorded by smart meters with built-in AD converters. Specifically, interpolation is used to align all data streams. For data collected at different frequencies, such as temperature, which is usually once a minute, and current and voltage, which are several times per second, linear interpolation will be used to resample the temperature data to align it with the current and voltage data timestamps. For any time point t', the interpolation of current data is expressed as: Among them, t and t+Δt are known data points, and t' is the time point that needs to be interpolated. In the specific implementation, the above preprocessing is signal filtering, including: low-pass filtering is represented by implementing a fourth-order Butterworth low-pass filter for current and voltage signals, with a cutoff frequency set to 50Hz to remove high-frequency noise above the grid frequency. The filter formula is: Among them, ωc =2π×50Hz, n=4. The smoothing process is to apply sliding window average filtering to the temperature data. The window size is set to 5 minutes, corresponding to 5 data points. The smoothed temperature is: Then, data quality control and anomaly detection are performed. Specifically, outlier detection is performed on all data streams. The moving average μ and standard deviation σ of each data type are calculated, and data points outside the range of μ±3σ are dynamically identified and marked as abnormal. Error and warning messages in communication logs are automatically analyzed, combined with abnormal fluctuations in current, voltage, and temperature data, to identify potential equipment failures or data transmission problems.

[0071] The detection unit 20 is used to extract key features of the plurality of transmission data, determine a dynamic threshold value according to the key features, and perform anomaly detection to obtain a second anomaly detection result, wherein the key features are features characterizing a data transmission failure, and the dynamic threshold value is used to determine whether the transmission data has a failure;

[0072] Specifically, in the specific implementation, this step can automatically learn and extract key features that are helpful for fault diagnosis from unlabeled data. This learning mode does not rely on pre-defined labels, reduces the need for manual input, and improves the ability to handle unknown fault modes. Through devices such as contrast learning, the system can automatically mark potential faults, significantly improve the sensitivity and accuracy of fault detection, and solve the shortcomings of traditional devices in identifying complex faults. Among them, the empirical mode decomposition (EMD) device is used to decompose the time series data I(t) and V(t) into a series of intrinsic mode functions (IMFs) and a residual term. The device extracts the oscillation mode of the signal step by step through the interpolation calculation of local extreme values ​​to adapt to the characteristics of nonlinear and non-stationary signals. In the specific implementation, the present invention also provides a comprehensive data management device for data marking and recording. Specifically, for each detected anomaly, detailed time information is recorded, including the exact start time and end time of the anomaly, and the duration. At the same time, all key information related to the anomaly is recorded, including spectral entropy values, selected IMFs, environmental conditions (such as temperature, humidity) when the anomaly occurs, and equipment status. Finally, a real-time updated data dashboard is provided to display current and historical abnormal events, including its spectrum entropy change trend graph, affected power lines and equipment. Developing APIs allows the system to be integrated with other fault management and monitoring systems to achieve instant data sharing and multi-system collaboration.

[0073] The generation unit 30 is used to use a causal network to analyze the above-mentioned first abnormality detection result and the above-mentioned second abnormality detection result, determine the initial position of the data transmission failure of the above-mentioned distribution network terminal, implement federated learning between multiple nodes in the above-mentioned causal network, and further determine the final position of the data transmission failure of the above-mentioned distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning, and generate a fault repair strategy according to the above-mentioned final position to repair the data transmission failure of the above-mentioned distribution network terminal.

[0074] Specifically, in the specific implementation, this step applies causal inference technology to analyze the results of anomaly detection and determine the specific cause and precise location of the fault. This graph-based causal inference algorithm can accurately identify the source of the fault and the propagation path in the complex system of the power grid. It solves the problem that traditional devices only rely on correlation analysis and cannot accurately locate and diagnose the cause of the fault, thereby significantly improving the efficiency and effectiveness of fault handling.

[0075] Through this embodiment, a variety of transmission data of the distribution network terminal is obtained, and the transmission data is log analyzed to obtain a first abnormal detection result; the key features of the various transmission data are extracted, and the dynamic threshold is determined according to the key features, and an abnormal detection is performed to obtain a second abnormal detection result, wherein the key feature is a feature that characterizes the data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure; the first abnormal detection result and the second abnormal detection result are analyzed using a causal network to determine the initial position of the data transmission failure of the distribution network terminal, and federated learning is implemented between multiple nodes in the causal network. By adjusting the weight distribution and parameter update strategy in the federated learning, the final position of the data transmission failure of the distribution network terminal is further determined, and a fault repair strategy is generated according to the final position to repair the data transmission failure of the distribution network terminal. Compared with the prior art, which is difficult to identify the cause of the distribution network data transmission failure, this method significantly improves the diagnostic ability of complex and unknown fault modes by automatically learning and extracting key features, and can accurately distinguish the causal relationship and simple correlation between the fault phenomena. By using the federated learning framework, multiple distribution terminals can share learning results under the premise of ensuring data privacy, thereby accurately locating the cause of the failure. Therefore, the problem that it is difficult to identify the cause of the distribution network data transmission failure in the prior art can be solved.

[0076] In the specific implementation process, the above-mentioned generation unit includes a first determination module, a construction module and a second determination module. The first determination module is used to obtain the time series data of the sensor of the above-mentioned distribution network terminal and synchronize the timestamp, define each of the above-mentioned sensors as a node, and determine the connection relationship between the multiple above-mentioned nodes according to the physical connection relationship of the power grid and the liquidity of historical data; the construction module is used to use the transfer entropy device Calculate the causal relationship between the multiple nodes to obtain the transfer entropy value, and construct a directed graph based on the transfer entropy value, wherein the edge of the directed graph is determined by the size of the transfer entropy value and represents the information flow or influence from one node to another. represents the state of the node at time t, represents the state of the edge at time t, represents the state of the above edge at time t+1, TE i→j represents the above transfer entropy value; the second determination module is used to determine the node where the abnormality first occurs as the initial starting point of the data transmission fault, and use a graph search algorithm in the above directed graph to determine multiple fault propagation paths from the above initial starting point to obtain multiple above initial positions. The device determines the initial position of the fault through the above steps, so that the position of the fault can be further located based on the initial position.

[0077] Specifically, this step includes: collecting multivariate time series data of key nodes, such as current, voltage and environmental parameters (such as temperature), and ensuring the temporal consistency of all data. Specifically, multifunctional sensors are installed at key nodes of the distribution network, which can measure parameters such as current, voltage and ambient temperature, and can send data in real time. The global positioning system (GPS) or network time protocol (NTP) is used to ensure the synchronization of timestamps of all sensors so as to accurately capture and analyze the temporal relationship in the data. A set of safe and reliable data transmission systems is established to transmit the data collected by the sensors to a central database in real time. Encrypted transmission is used to ensure the security of data during transmission. A dynamic causal network is constructed based on the collected data to understand the dependencies and causal relationships between different nodes. For the present invention, each sensor or power grid component is defined as a node in the graph, and the connection relationship between nodes is defined according to the physical connection of the power grid and the mobility of historical data. The causal relationship between nodes is calculated using a transfer entropy device. The transfer drop not only measures the flow of information between the two systems, but also reflects the dynamic dependency between them. Identify the starting point of the fault and infer the propagation path of the fault in the network. Specifically, based on the anomaly detection results (data analysis from the previous step), identify the node where the anomaly first occurs as the possible starting point of the fault. Using the constructed causal graph, a graph search algorithm (such as depth-first search DFS) is used to trace the possible fault propagation path starting from the fault starting point. The path selection is based on the size of the transfer entropy value. A high transfer entropy value indicates a large information flow or influence, and a higher possibility of fault propagation. Generate a detailed fault report based on the analysis results, and use the discovered information to optimize the system's monitoring and diagnosis process. Specifically, record each identified fault in detail, including the fault type, affected nodes, time of fault occurrence, and inferred fault cause and consequences. According to the results of fault handling and subsequent operation and maintenance feedback, adjust the data acquisition parameters and causal analysis model to improve the accuracy of future fault prediction and location. Implement continuous system performance monitoring to evaluate the effectiveness of optimization measures and further adjust strategies.

[0078] In some optional embodiments, the detection unit includes a decomposition module and a reconstruction module, and the decomposition module is used to use the empirical mode decomposition method EMD (X (t)) = {IMF 1 ,IMF 2 ,......,IMF n ,R} decompose the above transmission data to obtain multiple intrinsic modes and a residual term, among which, IMF i represents the i-th intrinsic mode, R represents the residual term, and X(t) represents the input signal; the reconstruction module is used to reconstruct the signal according to the intrinsic mode. To filter high frequency noise and extract the above key features of the above transmission data, where k represents the number of the above natural modes, X rec (t) represents the reconstructed signal. The device extracts key features through the above steps, so that the fault position can be located through the key features.

[0079] In the specific implementation process, this step can automatically learn from unlabeled data and extract key features that are helpful for fault diagnosis. This learning mode does not rely on pre-defined labels, reduces the need for manual input, and improves the ability to handle unknown fault modes. Through devices such as comparative learning, the system can automatically mark potential faults, significantly improve the sensitivity and accuracy of fault detection, and solve the shortcomings of traditional devices in identifying complex faults. Among them, the empirical mode decomposition (EMD) device is used to decompose the time series data I(t) and V(t) into a series of intrinsic mode functions (IMFs) and a residual term. The device extracts the oscillation mode of the signal step by step through the interpolation calculation of local extreme values ​​to adapt to the characteristics of nonlinear and non-stationary signals.

[0080] In some other optional implementations, the detection unit includes a first calculation module, a second calculation module and a third determination module, the first calculation module is used to calculate the power spectrum of the reconstructed signal through a fast Fourier transform device, and the spectral entropy of the reconstructed signal is obtained as follows: Where p(f) is the normalized power spectral density, F represents the total number of frequency components, SE represents the above spectral entropy, and the above reconstructed signal is a signal obtained by performing empirical mode decomposition and signal reconstruction on the above key features; the second calculation module is used to calculate the moving average threshold and moving standard deviation threshold of the above key features according to the above spectral entropy: Among them, α represents the first smoothing factor, β represents the second smoothing factor, the first smoothing factor and the second smoothing factor are used for the weight of historical data on the current estimate, t represents time, μ SE represents the above moving average threshold, σ SE Represents the above moving standard deviation threshold; the determination module is used to determine whether the current spectral entropy meets the conditions In the case of, it is determined that the above transmission data is abnormal, among which, SE t The device detects through the dynamic threshold of the above steps, so that anomaly detection can be performed to accurately detect the abnormal position.

[0081] The specific implementation process also includes automatically adjusting the values ​​of α and β to optimize the sensitivity and accuracy of the thresholds based on the accuracy and feedback of past anomaly detection (such as actual fault confirmation). After diagnosis, the weight of historical data is adjusted based on the feedback of actual fault diagnosis results to improve the accuracy of future predictions.

[0082] In some optional implementations, the generation unit includes a third calculation module, an adjustment module, and an update module. The third calculation module is used to obtain the quality and reliability of the historical data of the node, and calculate the score Score according to the quality of the historical data and the reliability. i =a×Completeness i +b×Synchrony i , the node with the highest calculation score is used for training the federated learning model, where Score i represents the score of node i, a represents the quality coefficient of historical data, b represents the reliability coefficient, and completeness i Indicates the quality of the above historical data of the above node i, Synchrony i Represents the reliability of the node i; the adjustment module is used to adjust the learning rate η of the federated learning model new =η old ×exp(-γ×ΔL), where η old represents the above learning rate before adjustment, η new represents the adjusted learning rate, γ represents the adjustment factor of the learning rate, and ΔL represents the loss reduction of two consecutive iterations; the update module is used to aggregate the model parameters of each of the above nodes using an adaptive weighted average device To update the above federated learning model, where w i represents the data quality of the above node i, θ i represents the model parameters of the above node i, N represents the total number of the above nodes, θ global The device is adjusted and updated through the above steps, so that the fault location can be determined more accurately.

[0083] In the specific implementation process, this step specifically includes: adaptive initialization and node selection. Specifically, the node is scored according to the quality and reliability of its historical data, and nodes with high scores are selected to participate in model training. The following scoring formula is used to consider the data integrity and time synchronization of each node. The model parameters are initialized on the central server, and the initial parameter allocation is adjusted according to the node's score to ensure that high-quality nodes have a greater impact on the model. Each node trains the model based on local data and optimizes the learning process through an adaptive algorithm. Specifically, each node uses its local data for model training and optimizes the following loss function: Among them, D local is the local dataset of the node, loss is the loss function, and λ is the regularization parameter. A dynamic adjustment mechanism is introduced to adjust the learning rate according to the performance of the model on the local dataset: η new =η old×exp(-γ×ΔL), where η old and η new is the learning rate before and after adjustment, γ is the learning rate adjustment factor, ΔL is the loss reduction between two consecutive iterations, and the model parameters of each node are aggregated on the central server, and the global model is updated. Specifically, the model parameters of each node are aggregated using an adaptive weighted average device: Among them, w i is a weight dynamically calculated based on the data quality and model performance of node i, θ i is the model parameter of node i. The aggregated global model parameters are sent back to each node for the next round of training. The weight w of the node based on the performance of the global model i Make adjustments to optimize subsequent aggregation effects. Monitor the performance of the federated learning process and optimize the learning strategy and parameters based on the monitoring results. Specifically, regularly evaluate the performance of the global model on independent test sets, including the accuracy and recall of fault diagnosis. Analyze the performance of the model on different types of fault diagnosis tasks and identify optimization space. Further, adjust parameters such as α, β, and γ based on the performance evaluation results to improve learning efficiency and model accuracy. Update data collection and preprocessing strategies to improve data quality and further strengthen the foundation of model learning.

[0084] In some optional implementations, the generation unit includes an acquisition module and a fourth calculation module. The acquisition module is used to acquire real-time current data and real-time voltage data, apply a time series anomaly detection algorithm to identify potential fault signals in real time, and use a machine learning classification model to automatically classify the fault types of the potential fault signals; the fourth calculation module is used to construct an influence propagation model based on graph theory, which is used to simulate the influence diffusion process starting from the fault point, determine a priority algorithm based on the fault impact and repair difficulty according to the above influence diffusion process, and calculate the above fault repair strategy according to the above priority algorithm. The device generates a fault repair strategy through the above steps, so that the fault can be automatically repaired after locating the fault.

[0085] Specifically, this step includes: 1. Real-time identification of faults in the distribution network, and intelligent classification based on fault characteristics, so as to take targeted response measures. Among them, fault signal identification uses the current and voltage data monitored in real time, and applies the time series anomaly detection algorithm to identify potential fault signals in real time. The fault detection formula is based on the physical laws of electrical systems, such as Ohm's law and Kirchhoff's current law, which are used to calculate node current and voltage deviations: Among them I expected and V expected is the expected value based on network load prediction, I measure d and V measured is the measured value. Intelligent classification of fault types is to automatically classify fault types using machine learning classification models, such as short circuit, open circuit, equipment failure, etc. Model training uses historical fault data and corresponding response effects to enhance classification accuracy. Second, evaluate the impact of faults on distribution network operations, reasonably allocate response priorities, and ensure power supply safety in key areas and key users. Specifically, develop an impact propagation model based on graph theory, which can simulate the impact diffusion process starting from the fault point; wherein, the impact propagation model formula may include: S(t+1)=S(t)·A+B·E(t), where S(t) is the system state vector at time t, A is the adjacency matrix of the network topology, B is the impact diffusion coefficient matrix, and E(t) is the fault impact vector. Further, design a priority algorithm based on fault impact and repair difficulty, which considers both the breadth and depth of fault impact. Priority calculation may include: P=α·Impact+β·Difficulty, where α and β are weight parameters, Impact is the fault impact score, and Difficulty is the repair difficulty score. 3. Automatic resource allocation and fault repair task dispatching, optimize the utilization of human and material resources, and shorten the troubleshooting and recovery time. Specifically, the optimization algorithm is used to automatically determine the best resource allocation plan, including the scheduling of the repair team and the allocation of necessary equipment. The resource optimization model may use linear programming or integer programming techniques to minimize the total cost or total response time. For the present invention, the automatic dispatching of repair tasks is to automatically dispatch repair tasks to the most suitable maintenance team in combination with the fault type and the geographic information system (GIS). The task dispatch algorithm takes into account the team's skills, availability, and distance from the fault point to optimize the response time and repair efficiency.

[0086] In the specific implementation process, an adaptive learning mechanism is introduced to build a real-time adaptive fault diagnosis and learning framework, which automatically adjusts the fault detection and repair strategy according to multi-source data and external environmental changes. In the specific implementation, the system automatically adjusts the fault detection and response strategy according to performance feedback and external environmental changes through an adaptive learning mechanism. This ensures that the system can continuously adapt to new fault modes and environmental changes, solving the problem that traditional systems cannot optimize and update themselves.

[0087] Choose a suitable online learning algorithm as the basis, such as online stochastic gradient descent (SGD) or online recursive least squares estimation (RLS). Design an adaptive learning rate adjustment mechanism to dynamically adjust the learning rate through online performance feedback: Among them, η t is the learning rate at time t, γ is the adjustment factor, is the gradient of the loss function ξ with respect to the parameter θ. Every time new data is received, the model parameters are updated immediately using these data: Among them, xt and t Represent the input data and observed output at time t, respectively, and f is the fault detection model. Furthermore, the impact of the fault on the distribution network is evaluated in real time, and resource allocation is optimized to minimize the negative impact of the fault.

[0088] Specifically, the graph theory model is used to evaluate the possible paths and impact range of fault spread in real time, and the impact degree is calculated in combination with network flow theory: Among them, c n is the importance coefficient of node n, Flow disrupted (n,t) is the amount of traffic interruption due to the fault at time t.

[0089] Then, according to the real-time evaluation results and pre-defined priority rules, technicians and maintenance equipment are dynamically allocated: Allocate(t) = argmin r∈R {Cost(r)+λ·TimeDelay(r,t)}, where R is the set of available resources, Cost(r) and TimeDelay(r,t) represent the cost and possible time delay of using resource r, respectively, and λ is a trade-off factor.

[0090] Furthermore, the system performance is continuously monitored, and the learning strategy and system configuration are adjusted based on performance feedback to ensure that the system is continuously optimized and adapts to grid changes. Among them, a comprehensive monitoring system is implemented to regularly collect data on fault detection accuracy, response time, and user satisfaction. The monitoring data is analyzed using data-driven devices to identify performance bottlenecks and optimization opportunities. Based on the monitoring results, model parameters, learning strategies, and resource allocation schemes are automatically adjusted to continuously improve system efficiency and accuracy.

[0091] In some optional implementations, after acquiring multiple transmission data of the distribution network terminal, the above-mentioned device further includes a filtering unit and a smoothing processing unit, wherein the filtering unit is used to perform low-pass filtering on the above-mentioned transmission data through a fourth-order Butterworth low-pass filter to obtain low-pass filtered data to remove high-frequency noise greater than the power grid frequency; and the smoothing processing unit is used to apply a sliding window to smooth the above-mentioned low-pass filtered data to obtain the above-mentioned pre-processed transmission data. The device performs data pre-processing through the above steps, so that the transmission data can be processed into a unified format that is easy to process, so as to facilitate subsequent data processing.

[0092] Specifically, the above preprocessing is signal filtering, including: low-pass filtering is represented by implementing a fourth-order Butterworth low-pass filter for the current and voltage signals, with a cutoff frequency set to 50 Hz to remove high-frequency noise above the grid frequency. The filter formula is: Among them, ω c=2π×50Hz, n=4. The smoothing process is to apply sliding window average filtering to the temperature data. The window size is set to 5 minutes, corresponding to 5 data points. The smoothed temperature is: Then perform data quality control and anomaly detection.

[0093] Specifically, outlier detection is implemented for all data streams. The moving average μ and standard deviation σ of each data type are calculated, and data points outside the range of μ+3σ are dynamically identified and marked as abnormal. Error and warning messages in communication logs are automatically analyzed, combined with abnormal fluctuations in current, voltage, and temperature data, to identify potential equipment failures or data transmission issues.

[0094] The above-mentioned processing device for data transmission failure of distribution network terminal includes a processor and a memory. The above-mentioned analysis unit, detection unit and generation unit are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions. The above-mentioned modules are all located in the same processor; or, the above-mentioned modules are located in different processors in any combination.

[0095] The processor includes a kernel, and the kernel retrieves the corresponding program unit from the memory. One or more kernels can be set, and the problem of difficulty in identifying the location of the distribution network data transmission fault in the prior art can be solved by adjusting the kernel parameters.

[0096] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0097] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the method for processing data transmission failure of the distribution network terminal.

[0098] Specifically, the method for handling the data transmission failure of the distribution network terminal includes:

[0099] Step S201: Acquire various transmission data of the distribution network terminal, perform log analysis on the transmission data, and obtain a first abnormality detection result;

[0100] Step S202: extracting key features of the plurality of transmission data, determining a dynamic threshold according to the key features, and performing anomaly detection to obtain a second anomaly detection result, wherein the key features are features characterizing a data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure;

[0101] Step S203: Use a causal network to analyze the first anomaly detection result and the second anomaly detection result, determine the initial position of the data transmission failure of the distribution network terminal, implement federated learning between multiple nodes in the causal network, and further determine the final position of the data transmission failure of the distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning, and generate a fault repair strategy based on the final position to repair the data transmission failure of the distribution network terminal.

[0102] An embodiment of the present invention provides an electronic device, including a processor, a memory, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, at least the following steps are implemented:

[0103] Step S201: Acquire various transmission data of the distribution network terminal, perform log analysis on the transmission data, and obtain a first abnormality detection result;

[0104] Step S202: extracting key features of the plurality of transmission data, determining a dynamic threshold according to the key features, and performing anomaly detection to obtain a second anomaly detection result, wherein the key features are features characterizing a data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure;

[0105] Step S203: Use a causal network to analyze the first anomaly detection result and the second anomaly detection result, determine the initial position of the data transmission failure of the distribution network terminal, implement federated learning between multiple nodes in the causal network, and further determine the final position of the data transmission failure of the distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning, and generate a fault repair strategy based on the final position to repair the data transmission failure of the distribution network terminal.

[0106] The present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps of the above method in each embodiment of the present application are implemented:

[0107] Step S201: Acquire various transmission data of the distribution network terminal, perform log analysis on the transmission data, and obtain a first abnormality detection result;

[0108] Step S202: extracting key features of the plurality of transmission data, determining a dynamic threshold according to the key features, and performing anomaly detection to obtain a second anomaly detection result, wherein the key features are features characterizing a data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure;

[0109] Step S203: Use a causal network to analyze the first anomaly detection result and the second anomaly detection result, determine the initial position of the data transmission failure of the distribution network terminal, implement federated learning between multiple nodes in the causal network, and further determine the final position of the data transmission failure of the distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning, and generate a fault repair strategy based on the final position to repair the data transmission failure of the distribution network terminal.

[0110] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0111] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0112] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0113] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0114] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0115] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0116] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0117] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0118] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0119] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0120] 1) In the processing method of the data transmission failure of the distribution network terminal of the present application, a plurality of transmission data of the distribution network terminal are obtained, and the transmission data are log analyzed to obtain a first abnormal detection result; the key features of the plurality of transmission data are extracted, and the dynamic threshold is determined according to the key features, and an abnormal detection is performed to obtain a second abnormal detection result, wherein the key feature is a feature characterizing the data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure; the first abnormal detection result and the second abnormal detection result are analyzed using a causal network to determine the initial position of the data transmission failure of the distribution network terminal, and federated learning is implemented between multiple nodes in the causal network, and the final position of the data transmission failure of the distribution network terminal is further determined by adjusting the weight distribution and parameter update strategy in the federated learning, and a fault repair strategy is generated according to the final position to repair the data transmission failure of the distribution network terminal. Compared with the prior art, which is difficult to identify the cause of the data transmission failure of the distribution network, the diagnostic ability of complex and unknown fault modes is significantly improved by automatic learning and extraction of key features, and the causal relationship and simple correlation between the fault phenomena can be accurately distinguished. By using the federated learning framework, multiple distribution terminals can share learning results under the premise of ensuring data privacy, thereby accurately locating the cause of the fault. Therefore, the problem in the prior art that it is difficult to identify the cause of a data transmission failure in a distribution network can be solved.

[0121] 2) In the processing device for data transmission failure of the distribution network terminal of the present application, a plurality of transmission data of the distribution network terminal are obtained, and the transmission data are log analyzed to obtain a first abnormal detection result; the key features of the plurality of transmission data are extracted, and the dynamic threshold is determined according to the key features, and an abnormal detection is performed to obtain a second abnormal detection result, wherein the key feature is a feature characterizing the data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure; the first abnormal detection result and the second abnormal detection result are analyzed using a causal network to determine the initial position of the data transmission failure of the distribution network terminal, and federated learning is implemented between multiple nodes in the causal network, and the final position of the data transmission failure of the distribution network terminal is further determined by adjusting the weight distribution and parameter update strategy in the federated learning, and a fault repair strategy is generated according to the final position to repair the data transmission failure of the distribution network terminal. Compared with the prior art, which is difficult to identify the cause of the data transmission failure of the distribution network, the diagnostic ability of complex and unknown fault modes is significantly improved by automatic learning and extraction of key features, and the causal relationship and simple correlation between the fault phenomena can be accurately distinguished. By using the federated learning framework, multiple distribution terminals can share learning results under the premise of ensuring data privacy, thereby accurately locating the cause of the fault. Therefore, the problem in the prior art that it is difficult to identify the cause of a data transmission failure in a distribution network can be solved.

[0122] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for processing data transmission failure of a distribution network terminal, characterized in that: include: Acquire multiple transmission data of the distribution network terminal, perform log analysis on the transmission data, and obtain a first abnormality detection result; Extracting multiple key features of the transmission data, determining a dynamic threshold according to the key features, and performing anomaly detection to obtain a second anomaly detection result, wherein the key feature is a feature characterizing a data transmission failure, and the dynamic threshold is used to determine whether the transmission data has a failure; The first anomaly detection result and the second anomaly detection result are analyzed using a causal network to determine the initial location of the data transmission failure of the distribution network terminal, and federated learning is implemented between multiple nodes in the causal network. The final location of the data transmission failure of the distribution network terminal is further determined by adjusting the weight distribution and parameter update strategy in the federated learning, and a fault repair strategy is generated according to the final location to repair the data transmission failure of the distribution network terminal.

2. The processing method according to claim 1, characterized in that: Analyzing the first abnormality detection result and the second abnormality detection result using a causal network to determine an initial location of the data transmission fault of the distribution network terminal includes: Acquire the time series data of the sensor of the distribution network terminal and synchronize the timestamp, define each sensor as a node, and determine the connection relationship between multiple nodes according to the physical connection relationship of the power grid and the liquidity of historical data; Using the transfer entropy method Calculating the causal relationship between the plurality of nodes to obtain a transfer entropy value, and constructing a directed graph according to the transfer entropy value, wherein the edge of the directed graph is determined by the size of the transfer entropy value and represents the information flow or influence from one node to another node, represents the state of the node at time t, represents the state of the edge at time t, represents the state of the edge at time t+1, TE i→j represents the transfer entropy value; The node where the abnormality first occurs is determined as the initial starting point of the data transmission failure, and a graph search algorithm is used in the directed graph to determine multiple fault propagation paths starting from the initial starting point to obtain multiple initial positions.

3. The processing method according to claim 1, characterized in that: Extract a variety of key features of the transmission data, including: Using the empirical mode decomposition method EMD(X(t))={IMF1,IMF2,......,IMF n ,R} decompose the transmission data to obtain multiple intrinsic modes and a residual term, where IMF i represents the i-th natural mode, R represents the residual term, and X(t) represents the input signal; Signal reconstruction is performed based on the natural modes To filter high frequency noise and extract the key features of the transmission data, where k represents the number of the natural modes, X rec (t) represents the reconstructed signal.

4. The processing method according to claim 1, characterized in that: Determining a dynamic threshold according to the key feature and performing anomaly detection to obtain a second anomaly detection result includes: The power spectrum of the reconstructed signal is calculated by the fast Fourier transform method, and the spectral entropy of the reconstructed signal is obtained as follows: Where p(f) is the normalized power spectral density, F represents the total number of frequency components, SE represents the spectral entropy, and the reconstructed signal is a signal obtained by performing empirical mode decomposition and signal reconstruction on the key feature; The moving average threshold and moving standard deviation threshold of the key feature are calculated according to the spectral entropy: Wherein, α represents the first smoothing factor, β represents the second smoothing factor, the first smoothing factor and the second smoothing factor are used for the weight of historical data on the current estimate, t represents time, μ SE represents the moving average threshold, σ SE represents the moving standard deviation threshold; The current spectral entropy meets the conditions In the case of, determining that the transmission data is abnormal, wherein SE t Represents the current spectral entropy corresponding to time t.

5. The processing method according to claim 1, characterized in that: Implementing federated learning among multiple nodes in the causal network, and further determining the final location of the data transmission failure of the distribution network terminal by adjusting the weight distribution and parameter update strategy in the federated learning, including: Obtain the historical data quality and reliability of the node, and calculate the score based on the historical data quality and the reliability i =a×Completeness i +b×Synchrony i , the node with the highest calculation score is used for training the federated learning model, where Score i represents the score of node i, a represents the quality coefficient of historical data, b represents the reliability coefficient, and completeness i represents the quality of the historical data of the node i, Synchrony i represents the reliability of the node i; Adjust the learning rate η of the federated learning model new =η old ×exp(-γ×ΔL), where η old represents the learning rate before adjustment, η new represents the adjusted learning rate, γ represents the adjustment factor of the learning rate, and ΔL represents the loss reduction between two consecutive iterations; Adopt adaptive weighted average method to aggregate the model parameters of each node To update the federated learning model, where w i represents the data quality of the node i, θ i represents the model parameters of the node i, N represents the total number of the nodes, θ global represents the model parameters after aggregation.

6. The processing method according to claim 1, characterized in that: Generate a fault repair strategy according to the final position, including: Acquire real-time current data and real-time voltage data, apply a time series anomaly detection algorithm to identify potential fault signals in real time, and use a machine learning classification model to automatically classify the fault type of the potential fault signal; An impact propagation model based on graph theory is constructed to simulate the impact diffusion process starting from the fault point, a priority algorithm based on the fault impact and repair difficulty is determined according to the impact diffusion process, and the fault repair strategy is calculated according to the priority algorithm.

7. The processing method according to claim 1, characterized in that: After acquiring the various transmission data of the distribution network terminal, the method further includes: The transmission data is low-pass filtered by a fourth-order Butterworth low-pass filter to obtain low-pass filtered data, so as to remove high-frequency noise greater than the power grid frequency; The low-pass filtered data is smoothed using a sliding window to obtain the pre-processed transmission data.

8. A device for processing data transmission failure of a distribution network terminal, characterized in that: include: An analysis unit, configured to obtain a plurality of transmission data of the distribution network terminal, perform log analysis on the transmission data, and obtain a first abnormality detection result; a detection unit, configured to extract key features of the plurality of transmission data, determine a dynamic threshold value according to the key features, and perform anomaly detection to obtain a second anomaly detection result, wherein the key features are features characterizing a data transmission failure, and the dynamic threshold value is used to determine whether the transmission data has a failure; A generating unit is used to analyze the first anomaly detection result and the second anomaly detection result by using a causal network, determine the initial position of the data transmission fault of the distribution network terminal, implement federated learning among multiple nodes in the causal network, further determine the final position of the data transmission fault of the distribution network terminal by adjusting the weight distribution and parameter updating strategy in the federated learning, and generate a fault repair strategy according to the final position to repair the data transmission fault of the distribution network terminal.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for processing data transmission failure of a distribution network terminal according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for processing distribution network terminal data transmission failures as described in any one of claims 1 to 7.

Citation Information

Cited By

  • Safety protection system for power distribution terminal

    CN119726576A

  • A safety protection system for a distribution terminal

    CN119726576B

  • Centralized-distributed cooperative distribution network fault detection method based on LOF-VAE-GPT

    CN119805053A

  • A centralized-distributed collaborative distribution network fault detection method based on LOF-VAE-GPT

    CN119805053B

  • Programmable high-real-time electric power communication platform

    CN120768760A