Distribution network fault monitoring method and system with edge calculation and AI intelligent identification

Abstract

The embodiment of the invention relates to the technical field of information, in particular to a distribution network fault monitoring method and system with edge calculation and AI intelligent identification. The method comprises the following steps: deploying an edge computing node at a transformer of a distribution network; periodically acquiring the acquired current monitoring data and voltage monitoring data; periodically acquiring oil sample monitoring data, working noise of the transformer and ultrasonic data acquired by a transformer shell; current features and voltage features are extracted, and a line fault prediction result is obtained; comparing the oil sample monitoring data with reference oil sample data to obtain an oil sample comparison result; extracting noise frequency domain features, and extracting ultrasonic frequency domain features; inputting the oil sample comparison result, the noise frequency domain characteristics and the ultrasonic frequency domain characteristics into a pre-configured fault diagnosis model to obtain a transformer fault prediction result; and obtaining a distribution network fault prediction result according to the line fault prediction result and the transformer fault prediction result.

Description

Technical Field

[0001] Multiple embodiments of this specification relate to the field of information technology, and specifically to a distribution network fault monitoring method and system with edge computing and AI intelligent identification. Background Art

[0002] With growing electricity demand and increasing grid complexity, traditional distribution network fault monitoring methods are no longer able to meet the reliability, real-time, and intelligent management requirements of modern power grids. Traditional methods primarily rely on scheduled maintenance and post-event repairs, which are not only inefficient but often fail to detect potential faults in a timely manner, leading to frequent power outages and severely impacting power quality and user satisfaction. In recent years, the development of edge computing technology and artificial intelligence (AI) has brought new solutions to distribution network fault monitoring. Edge computing processes data at the source, reducing data transmission latency and central server load, making it particularly suitable for scenarios requiring rapid response. However, existing distribution network monitoring systems combining edge computing and AI still face many challenges, including the processing of large data volumes and heterogeneous data from multiple sources. Therefore, further research on distribution network fault monitoring technologies is needed. Summary of the Invention

[0003] Multiple embodiments of this specification describe a distribution network fault monitoring method and system with edge computing and AI intelligent identification.

[0004] In a first aspect, an embodiment of this specification provides a distribution network fault monitoring method with edge computing and AI intelligent identification, characterized in that it includes the following steps:

[0005] Deploy edge computing nodes at transformers in the distribution network, and connect them to the central server.

[0006] The central server periodically acquires the collected current monitoring data and voltage monitoring data;

[0007] The edge computing node periodically acquires oil sample monitoring data, transformer operating noise, and ultrasonic data collected from the transformer casing;

[0008] The central server extracts features of the current monitoring data and the voltage monitoring data to obtain current features and voltage features, and obtains line fault prediction results based on the current features and voltage features;

[0009] The edge computing node compares the oil sample monitoring data with the reference oil sample data to obtain an oil sample comparison result;

[0010] The edge computing node extracts the frequency domain features of the working noise as noise frequency domain features, and extracts the frequency domain features of the ultrasound data as ultrasound frequency domain features;

[0011] Inputting the oil sample comparison results, noise frequency domain characteristics and ultrasonic frequency domain characteristics into a pre-configured fault diagnosis model to obtain a transformer fault prediction result;

[0012] A distribution network fault prediction result is obtained based on the line fault prediction result and the transformer fault prediction result.

[0013] In a second aspect, the embodiments of this specification provide a distribution network fault monitoring system with edge computing and AI intelligent identification, characterized by including:

[0014] Deployment module: deploy edge computing nodes at transformers in the distribution network, and connect the edge computing nodes to the central server;

[0015] The first acquisition module, the central server periodically acquires the collected current monitoring data and voltage monitoring data;

[0016] The second acquisition module, the edge computing node periodically acquires oil sample monitoring data, transformer operating noise, and ultrasonic data collected from the transformer casing;

[0017] A first diagnostic module, wherein the central server extracts features of the current monitoring data and the voltage monitoring data to obtain current features and voltage features, and obtains a line fault prediction result based on the current features and voltage features;

[0018] A comparison module, wherein the edge computing node compares the oil sample monitoring data with the reference oil sample data to obtain an oil sample comparison result;

[0019] A feature module, wherein the edge computing node extracts the frequency domain features of the working noise as noise frequency domain features, and extracts the frequency domain features of the ultrasound data as ultrasound frequency domain features;

[0020] The second diagnostic module inputs the oil sample comparison results, noise frequency domain characteristics and ultrasonic frequency domain characteristics into a pre-configured fault diagnosis model to obtain a transformer fault prediction result;

[0021] The summarizing module obtains the distribution network fault prediction result according to the line fault prediction result and the transformer fault prediction result.

[0022] In a third aspect, embodiments of this specification provide an electronic device, including a processor and a memory;

[0023] The processor is connected to the memory;

[0024] The memory is used to store executable program code;

[0025] The processor reads the executable program code stored in the memory to run a program corresponding to the executable program code, so as to execute the method described in any one of the above aspects.

[0026] In a fourth aspect, an embodiment of this specification provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method described in any one of the above aspects is implemented.

[0027] In a fifth aspect, embodiments of this specification provide a computer program product, including a computer program, which implements the method described in any of the above aspects when executed by a processor.

[0028] The beneficial effects of the technical solutions provided by some embodiments of this specification include at least:

[0029] In various embodiments of this specification, a distribution network fault monitoring method and system are provided. By integrating and analyzing multi-source heterogeneous data (gas composition in oil, operating noise spectrum, and casing ultrasonic signals), they can more comprehensively reflect the internal operating status of the transformer, addressing the information insufficiency of a single data source. By extracting noise and ultrasonic frequency domain features, they can capture subtle abnormal signals before a significant equipment failure occurs, thereby providing early warning. By utilizing an edge computing architecture to perform feature extraction and some inference tasks locally, they reduce the need to transmit raw data to the cloud and improve the real-time nature of fault prediction.

[0030] Other features and advantages of the various embodiments of this specification will be further disclosed in the following detailed description and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of this specification, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0032] Figure 1 This is a schematic diagram of the application of the distribution network fault monitoring method provided in the embodiments of this specification.

[0033] Figure 2 This is a flow chart of a distribution network fault monitoring method provided in an embodiment of this specification.

[0034] Figure 3 This is a schematic diagram of reference oil sample data provided in the embodiments of this specification.

[0035] Figure 4 This is a schematic diagram of the distribution network fault monitoring system provided in the embodiments of this specification.

[0036] Figure 5 This is a schematic diagram of an electronic device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0037] The following is an explanation and description of the technical solutions of the embodiments of this specification in conjunction with the drawings of the embodiments of this specification. However, the following embodiments are only preferred embodiments of this specification and are not exhaustive. Based on the embodiments in the implementation mode, other embodiments obtained by those skilled in the art without making any creative work are all within the scope of protection of this specification.

[0038] Throughout this specification, the claims, and the accompanying drawings, the terms "first," "second," "third," and the like are used to distinguish between different items, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may include other steps or elements inherent to the process, method, product, or apparatus.

[0039] In the following description, terms such as "inside", "outside", "up", "down", "left", "right", etc. that indicate directions or positional relationships are only used to facilitate the description of the embodiments and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, they should not be understood as limitations on this specification.

[0040] The data involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection of relevant data complies with the relevant laws, regulations and standards of relevant countries and regions.

[0041] Before introducing the technical solution in this specification, the application scenarios and related technologies of the technical solution are introduced.

[0042] The distribution network is an important part of the power system. Figure 1 , primarily consisting of a substation, distribution lines, transformers 21, switchgear, reactive power compensation devices (such as capacitors), and user-side electrical equipment. The substation is responsible for converting the transmission voltage into a suitable distribution voltage for end users. The distribution lines transmit the electricity to end users via overhead lines or cables, and transformers 21 further regulate the voltage to meet varying load requirements. Switchgear controls the connection and disconnection of circuits, while reactive power compensation devices improve grid efficiency by adjusting the power factor.

[0043] The main types of faults in the distribution network include short circuits, broken wires, ground faults, equipment aging or insulation failure, overloads, and harmonic interference. Short circuits are often caused by lightning strikes, tree branches touching conductors, or equipment insulation deterioration, resulting in abnormally high currents. Equipment aging or insulation failure can cause partial discharge or overheating, accelerating equipment damage. Overloads can be caused by sudden load increases or inadequate line 22 design, leading to elevated temperatures and even fires. In this manual, line 22 and distribution line have the same meaning.

[0044] Existing distribution network monitoring systems face numerous practical challenges. These systems rely on single-source sensor data (such as current and voltage monitoring) and lack the ability to integrate and analyze heterogeneous data from multiple sources (such as oil sample gas composition, noise spectrum, and ultrasonic signals), making it difficult to comprehensively assess equipment status. Furthermore, they often rely on threshold comparisons or simple statistical models, which have limited ability to identify complex fault modes (such as early insulation degradation and intermittent arcing) and are prone to false alarms or missed alarms. These issues hinder the accuracy and efficiency of distribution network fault monitoring and urgently require further research and optimization.

[0045] Since this application involves some professional terms, these professional terms will be introduced below.

[0046] Edge computing node 12

[0047] Edge computing nodes 12 are key components in a distributed computing architecture. Deployed close to data sources (such as sensors, devices, or user terminals), they perform real-time processing and analysis at the data generation site. Their functions include data processing and analysis, data caching and storage, network optimization, and load balancing. They offer low latency, high reliability, and data privacy protection, making them particularly suitable for scenarios with high real-time requirements.

[0048] Oil sample monitoring data

[0049] Oil sample monitoring data refers to the physical and chemical properties collected from the insulating oil in transformer 21, used to assess oil quality and equipment health. It is an important basis for early warning of transformer 21 faults, particularly in detecting insulation aging, partial discharge, or overheating. Common monitoring indicators include dissolved gas analysis (DGA), moisture content, acid value, viscosity, and dielectric strength. For example, DGA measures the content and ratio of gases such as hydrogen (H2), methane (CH4), and acetylene (C2H2) in the oil, and combines this with IEC 60599 standards (such as the Duval triangle method) to determine the type of fault (e.g., partial discharge, overheating).

[0050] Working noise

[0051] The operating noise of transformer 21 is the sum of mechanical vibration and airborne noise generated by the main structural design, cooling system operation, and electromagnetic forces. Its noise level is a key technical parameter for the design and operating status of transformer 21. Noise sources primarily include the magnetostrictive effect of the core, winding vibration, and mechanical noise from the cooling system. Monitoring is crucial for condition assessment and environmental management. Abnormal noise, such as sudden increases or changes in frequency, may indicate internal faults such as a loose core, winding deformation, or cooling system failure. For example, a transformer 21 experienced a significant increase in noise as the load increased. Analysis revealed localized deformation of the winding due to excessive electromagnetic forces, requiring immediate repair.

[0052] Ultrasound data

[0053] Ultrasonic data collected from the transformer 21 casing refers to mechanical vibration signals captured by ultrasonic sensors on the transformer 21 casing surface. These data are used to detect internal faults such as partial discharge and insulation degradation. The propagation characteristics of ultrasonic signals are closely related to the fault type and are an important basis for fault location and classification.

[0054] This manual provides a distribution network fault monitoring method with edge computing and AI intelligent identification. Figure 2 , including the steps of:

[0055] Step S1) Deploy edge computing nodes 12 at transformers 21 in the distribution network. Edge computing nodes 12 are connected to central servers 11. Deploying edge computing nodes 12 at transformers 21 and connecting them to central servers 11 is a key step in implementing intelligent monitoring and fault diagnosis for the distribution network. Edge nodes are installed in distribution rooms or ring main cabinets near transformers 21. Lightweight AI models (such as autoencoders and fault diagnosis models) are deployed on edge nodes.

[0056] In step S2, the central server 11 periodically acquires current and voltage monitoring data. A Hall-effect current sensor or Rogowski coil is used to perform non-contact current measurement of line 22. The output analog signal is amplified and conditioned before being fed into an analog-to-digital converter (ADC) for conversion into a digital signal. The high-voltage side voltage signal is collected using a resistor divider or a potential transformer (PT). The ADC performs digitization.

[0057] Step S3) The edge computing node 12 periodically obtains oil sample monitoring data, operating noise of the transformer 21, and ultrasonic data collected from the casing of the transformer 21.

[0058] Step S4) The central server 11 extracts features of the current monitoring data and the voltage monitoring data to obtain current features and voltage features, and obtains a fault prediction result of the line 22 based on the current features and the voltage features.

[0059] The method for extracting features of the current monitoring data and the voltage monitoring data includes:

[0060] Sampling the current monitoring data and the voltage monitoring data at a preset frequency to obtain a sampled current and a sampled voltage;

[0061] Establishing an auto-encoding model for the current monitoring data and the voltage monitoring data respectively, wherein the input of the auto-encoding model is the sampled current or the sampled voltage;

[0062] Obtaining a voltage feature extraction model and a current feature extraction model according to the left half of the trained autoencoder model;

[0063] The features of the current monitoring data and the voltage monitoring data are extracted according to the voltage feature extraction model and the current feature extraction model to obtain current features and voltage features.

[0064] The current and voltage signals are sampled synchronously at a preset frequency, such as 300Hz, to ensure timing consistency. The sampled data is converted into digital signals through the ADC and normalized, such as mapping the voltage value to the interval [0,1] to eliminate dimensional differences. A low-pass filter, such as the Butterworth filter, is used to remove high-frequency noise, or a wavelet transform is used to separate the signal from the noise. The autoencoder consists of an encoder and a decoder. The encoder compresses the input data into low-dimensional latent features, and the decoder attempts to reconstruct the original input. The optimization goal is to minimize the mean square error (MSE) between the input and the reconstructed output:

[0065]

[0066] in, is the original data, To reconstruct the data, Dropout (randomly discarding 20% ​​of neurons) is introduced in the encoder to prevent overfitting.

[0067] The method for obtaining line fault prediction results based on the current characteristics and voltage characteristics includes:

[0068] Fusing the current feature and the voltage feature to obtain a line state feature vector;

[0069] Inputting the line state feature vector into a pre-configured fault prediction model;

[0070] A line fault prediction result is obtained according to the response of the fault prediction model.

[0071] Accordingly, the method for fusing the current feature and the voltage feature includes:

[0072] aligning the current characteristics and the voltage characteristics along a time axis;

[0073] Extracting time series features of the current feature and the voltage feature respectively, wherein the time series features include a waveform distortion rate, a rate of change within a sliding window of a preset length, and a Pearson correlation coefficient;

[0074] constructing a mismatch feature vector according to a difference between the time series features of the current feature and the voltage feature;

[0075] The mismatching feature vector is used as the line state feature vector.

[0076] For example, the 10-dimensional features of the current feature vector extracted by the autoencoder and the 10-dimensional voltage feature vector are concatenated to form a 20-dimensional line state feature vector. Weights are assigned based on the sensitivity of the current and voltage features to the fault. For example, if a short-circuit fault is more dependent on current mutation characteristics, the current feature is given a higher weight. The time series features of the current and voltage are aligned, such as through a sliding window analysis with a sliding window length of 100 sampling points. The time domain correlation between the two, i.e., time series features, is extracted. Time series features include waveform distortion rate, rate of change within a sliding window of a preset length, and the Pearson correlation coefficient. A mismatch feature vector is constructed. The fault type (e.g., short circuit, overload, partial discharge) is associated with the mismatch feature vector, i.e., the line state feature vector. The waveform distortion rate, rate of change within a sliding window of a preset length, and the Pearson correlation coefficient are calculated using existing techniques. A machine learning model, such as a convolutional neural network (CNN) or a long short-term memory network (LSTM), is designed and trained using the line state feature vector associated with the fault label. Obtain a fault prediction model. For example, the model outputs a probability distribution of fault types (e.g., a Softmax output). The category with the highest probability is selected as the prediction result. A threshold is set based on the model output probability value (e.g., triggering an alarm when the confidence level is >70%) to avoid misjudgments.

[0077] Step S5) The edge computing node 12 compares the oil sample monitoring data with the reference oil sample data to obtain an oil sample comparison result.

[0078] Methods for comparing oil sample monitoring data with reference oil sample data to obtain oil sample comparison results include:

[0079] The oil sample monitoring data and the reference oil sample data are normalized and converted into vectors to obtain the oil sample monitoring vector and the reference oil sample vector;

[0080] A difference vector between the oil sample monitoring vector and the reference oil sample vector is calculated and used as the oil sample comparison result.

[0081] Please see the attached Figure 3 The oil sample monitoring data includes one or more parameters of transformer 21 oil, including moisture content, acid value, dielectric strength, dissolved gas content, oil color, and oil level. The reference oil sample data includes multiple oil sample data items, each of which records not only the aforementioned monitoring parameters but also the associated fault tag.

[0082] By analyzing oil sample monitoring data, faults such as insulating oil aging, internal partial discharge, overheating, arc discharge, and abnormal oil levels can be detected. Regarding insulating oil aging, due to long-term operation or the influence of high-temperature environments, the oil undergoes chemical changes, resulting in a decrease in its performance. This is specifically manifested as an increase in acid value and a decrease in dielectric strength. If timely measures are not taken, this situation will seriously affect the insulation performance of transformer 21 and increase the risk of failure. Internal partial discharge usually occurs when there are tiny defects or gaps inside transformer 21. This will cause the content of gases such as hydrogen and methane in the oil to increase significantly. Continuous partial discharge not only accelerates the aging process of the insulating material, but may also lead to more serious insulation breakdown accidents.

[0083] Overheating faults are generally caused by poor winding contact or a core short circuit, resulting in localized temperature increases. Abnormal concentrations of gases such as ethane and ethylene in the oil can occur. If overheating faults are not controlled, they may develop into more serious fault types. Arcing is a high-energy discharge phenomenon that commonly occurs in the event of a short circuit or severe insulation failure. A typical sign is a sharp increase in the acetylene gas content in the oil. Arcing is extremely destructive to equipment, and immediate action must be taken to prevent further damage. Abnormal oil levels occur when the oil level is too high or too low. A low oil level can affect the cooling effect of transformer 21 or reduce insulation performance; an excessively high oil level can cause seal damage due to expansion. In either case, the safe operation of transformer 21 is threatened. Oil sampling monitoring data can effectively identify various potential fault hazards in transformer 21, improving the speed and accuracy of fault diagnosis.

[0084] Step S6) The edge computing node 12 extracts frequency domain features of the operating noise, recording them as noise frequency domain features, and extracts frequency domain features of the ultrasound data, recording them as ultrasound frequency domain features. Frequency domain features can be extracted using publicly available techniques in the art, such as FFT and wavelet transform.

[0085] Step S7) The oil sample comparison result, noise frequency domain characteristics and ultrasonic frequency domain characteristics are input into a pre-configured fault diagnosis model to obtain a transformer 21 fault prediction result.

[0086] Methods for pre-configuring fault diagnosis models include:

[0087] Receiving the oil sample comparison result, the noise frequency domain characteristics, and the fault labels corresponding to the ultrasonic frequency domain characteristics respectively;

[0088] After the oil sample comparison result, noise frequency domain characteristics and ultrasonic frequency domain characteristics are represented by vectors, vector fusion is performed to obtain a feature vector;

[0089] The fault labels corresponding to the oil sample comparison result, the noise frequency domain characteristics, and the ultrasonic frequency domain characteristics are spliced ​​together as the labels of the feature vector;

[0090] Obtaining sample data according to the labeled feature vector;

[0091] A machine learning model is established and trained using the sample data, and a fault diagnosis model is obtained based on the trained machine learning model.

[0092] Edge nodes periodically collect oil sample comparison results, noise frequency domain characteristics, and ultrasonic frequency domain characteristics. After preprocessing, these characteristics are constructed into a uniformly formatted feature vector. This feature vector is then fed into a trained fault diagnosis model, which outputs a fault prediction for the current state, such as "loose core" or "normal operation." Once the fault type is determined, an alarm can be triggered and maintenance recommendations can be delivered. If the model is deployed at the edge, real-time feedback can also be provided to the local control system for rapid response.

[0093] Step S8) Obtain a distribution network fault prediction result based on the line fault prediction result and the transformer 21 fault prediction result. Combine the line fault prediction result and the transformer 21 fault prediction result as the distribution network fault prediction result.

[0094] On the other hand, this manual provides a distribution network fault monitoring system with edge computing and AI intelligent identification. Figure 4 ,include:

[0095] Deployment module 100, deploying edge computing nodes 12 at transformers 21 of the distribution network, and edge computing nodes 12 are connected to the central server 11;

[0096] The first acquisition module 200, the central server 11 periodically acquires the collected current monitoring data and voltage monitoring data;

[0097] The second acquisition module 300, the edge computing node 12 periodically acquires oil sample monitoring data, operating noise of the transformer 21, and ultrasonic data collected from the casing of the transformer 21;

[0098] In the first diagnostic module 400, the central server 11 extracts features of the current monitoring data and the voltage monitoring data to obtain current features and voltage features, and obtains a line fault prediction result based on the current features and voltage features;

[0099] Comparison module 500, the edge computing node 12 compares the oil sample monitoring data with the reference oil sample data to obtain an oil sample comparison result;

[0100] The feature module 600, the edge computing node 12 extracts the frequency domain features of the working noise and records them as noise frequency domain features, and extracts the frequency domain features of the ultrasound data and records them as ultrasound frequency domain features;

[0101] The second diagnostic module 700 inputs the oil sample comparison result, noise frequency domain characteristics and ultrasonic frequency domain characteristics into a pre-configured fault diagnosis model to obtain a transformer 21 fault prediction result;

[0102] The summarizing module 800 obtains a distribution network fault prediction result according to the line fault prediction result and the transformer 21 fault prediction result.

[0103] See also Figure 5 A schematic structural diagram of an electronic device provided in an embodiment of this specification is shown.

[0104] like Figure 5 As shown, the electronic device 1100 may include: at least one processor 1101, at least one network interface 1104, a user interface 1103, a memory 1105, and at least one communication bus 1102. The communication bus 1102 may be used to implement communication between the aforementioned components. The user interface 1103 may include buttons, and optionally may also include a standard wired interface or a wireless interface. The network interface 1104 may include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc. The processor 1101 may include one or more processing cores. The processor 1101 utilizes various interfaces and circuits 22 to connect the various components within the entire electronic device 1100. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 1105, and accessing data stored in the memory 1105, the processor 1101 performs various functions of the routing device 1100 and processes data. Optionally, the processor 1101 may be implemented in hardware using at least one of a DSP, an FPGA, and a PLA. The processor 1101 may integrate one or a combination of a CPU, a GPU, and a modem. The CPU primarily processes the operating system, user interface, and applications; the GPU is responsible for rendering and drawing content displayed on the display; and the modem handles wireless communications.

[0105] It is understandable that the above-mentioned modem may not be integrated into the processor 1101, but may be implemented by a separate chip.

[0106] Memory 1105 may include either RAM or ROM. Optionally, memory 1105 may include non-transitory computer-readable media. Memory 1105 may be used to store instructions, programs, codes, code sets, or instruction sets. Memory 1105 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, sound playback function, image playback function, etc.), instructions for implementing the aforementioned method embodiments, etc.; the data storage area may store data related to the aforementioned method embodiments, etc. Memory 1105 may also optionally be at least one storage device located remotely from the aforementioned processor 1101. Memory 1105, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and application programs. Processor 1101 may be configured to invoke the application programs stored in memory 1105 and execute the methods described in the aforementioned embodiments.

[0107] The embodiments of this specification also provide a computer-readable storage medium having instructions stored therein that, when executed on a computer or processor, cause the computer or processor to perform the steps of the aforementioned embodiments. If the components of the aforementioned electronic device are implemented as software functional units and sold or used as independent products, they may be stored in the computer-readable storage medium.

[0108] The embodiments of this specification also provide a computer program product, including a computer program, which implements multiple steps in the above embodiments when executed by a processor.

[0109] In the absence of conflict, the technical features in this embodiment and implementation scheme can be combined arbitrarily.

[0110] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product comprises multiple computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted via the computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium accessible by a computer or a data storage device such as a server or data center that integrates multiple available media. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state drive (SSD)).

[0111] When implemented via hardware or firmware, the aforementioned method flow is programmed into the hardware circuit to obtain the corresponding hardware circuit structure and realize the corresponding function. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit, whose logical function is determined by the user's device programming. Designers can "integrate" a digital system on a PLD through self-programming, eliminating the need for chip manufacturers to design and manufacture dedicated integrated circuit chips. Moreover, today, instead of manually manufacturing integrated circuit chips, this programming is often performed using "logic compiler" software. This is similar to the software compiler used in program development. Before compilation, the original code must also be written in a specific programming language, called a hardware description language (HDL). There are not just one HDL, but many. Those skilled in the art will also understand that simply by programming the method flow in one of the aforementioned hardware description languages ​​and programming it into the integrated circuit, a hardware circuit that implements the logical method flow can be easily obtained.

[0112] The embodiments described above are merely preferred embodiments of this specification and are not intended to limit the scope of this specification. Without departing from the design spirit of this specification, various modifications and improvements made to the technical solutions of this specification by ordinary technicians in this field should fall within the scope of protection determined by the claims of this specification.

Claims

1. A distribution network fault monitoring method with edge computing and AI intelligent identification, characterized in that: Including steps: Deploy edge computing nodes at transformers in the distribution network, and connect them to the central server. The central server periodically acquires the collected current monitoring data and voltage monitoring data; The edge computing node periodically acquires oil sample monitoring data, transformer operating noise, and ultrasonic data collected from the transformer casing; The central server extracts features of the current monitoring data and the voltage monitoring data to obtain current features and voltage features, and obtains line fault prediction results based on the current features and voltage features; The edge computing node compares the oil sample monitoring data with the reference oil sample data to obtain an oil sample comparison result; The edge computing node extracts the frequency domain features of the working noise as noise frequency domain features, and extracts the frequency domain features of the ultrasound data as ultrasound frequency domain features; Inputting the oil sample comparison results, noise frequency domain characteristics and ultrasonic frequency domain characteristics into a pre-configured fault diagnosis model to obtain a transformer fault prediction result; A distribution network fault prediction result is obtained based on the line fault prediction result and the transformer fault prediction result.

2. The distribution network fault monitoring method with edge computing and AI intelligent identification according to claim 1 is characterized in that: The method for extracting features of the current monitoring data and the voltage monitoring data includes: Sampling the current monitoring data and the voltage monitoring data at a preset frequency to obtain a sampled current and a sampled voltage; Establishing an auto-encoding model for the current monitoring data and the voltage monitoring data respectively, wherein the input of the auto-encoding model is the sampled current or the sampled voltage; Obtaining a voltage feature extraction model and a current feature extraction model according to the left half of the trained autoencoder model; The features of the current monitoring data and the voltage monitoring data are extracted according to the voltage feature extraction model and the current feature extraction model to obtain current features and voltage features.

3. The distribution network fault monitoring method with edge computing and AI intelligent identification according to claim 1 or 2, characterized in that: The method for obtaining a line fault prediction result based on the current characteristics and voltage characteristics includes: Fusing the current feature and the voltage feature to obtain a line state feature vector; Inputting the line state feature vector into a pre-configured fault prediction model; A line fault prediction result is obtained according to the response of the fault prediction model.

4. The distribution network fault monitoring method with edge computing and AI intelligent identification according to claim 3 is characterized in that: The method for fusing the current feature and the voltage feature includes: aligning the current characteristics and the voltage characteristics along a time axis; Extracting time series features of the current feature and the voltage feature respectively, wherein the time series features include a waveform distortion rate, a rate of change within a sliding window of a preset length, and a Pearson correlation coefficient; constructing a mismatch feature vector according to a difference between the time series features of the current feature and the voltage feature; The mismatching feature vector is used as the line state feature vector.

5. The distribution network fault monitoring method with edge computing and AI intelligent identification according to claim 1 or 2, characterized in that: Methods for comparing oil sample monitoring data with reference oil sample data to obtain oil sample comparison results include: The oil sample monitoring data and the reference oil sample data are normalized and converted into vectors to obtain the oil sample monitoring vector and the reference oil sample vector; A difference vector between the oil sample monitoring vector and the reference oil sample vector is calculated and used as the oil sample comparison result.

6. The distribution network fault monitoring method with edge computing and AI intelligent identification according to claim 1 or 2, characterized in that: Methods for pre-configuring fault diagnosis models include: Receiving the oil sample comparison result, the noise frequency domain characteristics, and the fault labels corresponding to the ultrasonic frequency domain characteristics respectively; After the oil sample comparison result, noise frequency domain characteristics and ultrasonic frequency domain characteristics are represented by vectors, vector fusion is performed to obtain a feature vector; The fault labels corresponding to the oil sample comparison result, the noise frequency domain characteristics and the ultrasonic frequency domain characteristics are spliced ​​together as the labels of the feature vector; Obtaining sample data according to the labeled feature vector; A machine learning model is established and trained using the sample data, and a fault diagnosis model is obtained based on the trained machine learning model.

7. A distribution network fault monitoring system with edge computing and AI intelligent identification, characterized by: include: Deployment module: deploy edge computing nodes at transformers in the distribution network, and connect the edge computing nodes to the central server; The first acquisition module, the central server periodically acquires the collected current monitoring data and voltage monitoring data; The second acquisition module, the edge computing node periodically acquires oil sample monitoring data, transformer operating noise, and ultrasonic data collected from the transformer casing; A first diagnostic module, wherein the central server extracts features of the current monitoring data and the voltage monitoring data to obtain current features and voltage features, and obtains a line fault prediction result based on the current features and voltage features; A comparison module, wherein the edge computing node compares the oil sample monitoring data with the reference oil sample data to obtain an oil sample comparison result; A feature module, wherein the edge computing node extracts the frequency domain features of the working noise as noise frequency domain features, and extracts the frequency domain features of the ultrasound data as ultrasound frequency domain features; The second diagnostic module inputs the oil sample comparison results, noise frequency domain characteristics and ultrasonic frequency domain characteristics into a pre-configured fault diagnosis model to obtain a transformer fault prediction result; The summarizing module obtains the distribution network fault prediction result according to the line fault prediction result and the transformer fault prediction result.

8. An electronic device, characterized in that: including a processor and a memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images