CN-122020290-A - Fault cause analysis method and system for small-current grounding system

CN122020290ACN 122020290 ACN122020290 ACN 122020290ACN-122020290-A

Abstract

A method for analyzing the fault cause of small-current grounding system includes such steps as obtaining the topological structure of small-current grounding fault system when the small-current grounding fault occurs, obtaining the equipment record of equipment whose physical distance from the fault point is nearest, extracting the zero-sequence voltage and current and the length of single cycle in equipment record, intercepting the cycle according to the length of single cycle and the set step length, processing the intercepted cycle, inputting the zero-sequence voltage, zero-sequence current and processed cycle to several discriminators for different fault causes to obtain the classification probability for judging different types of fault causes, and selecting final judging result based on the classification probability of different types of fault causes. The invention can meet the real-time requirement of the modern power distribution network on rapid fault diagnosis and isolation by means of the characteristic module and the classifier.

Inventors

ZHENG ZIMO
XU WEILI
DING RONGWEI
Li Ningqi
QI HAIXING
LIU WENXIANG
BI JUNJIE
CHEN CHENG

Assignees

北京丹华昊博电力科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260122

Claims (10)

1. The small current grounding system fault cause analysis method is characterized by comprising the following steps: when a small-current ground fault occurs, acquiring a topological structure of a small-current ground fault system; finding out a fault point in the topological structure, and acquiring equipment wave records of equipment with the closest physical distance to the fault point; extracting zero sequence voltage and zero sequence current in the equipment record and the length of a single cycle, performing cycle interception on the equipment record according to the length of the single cycle and a set step length, and processing the intercepted cycle; Inputting the zero sequence voltage, the zero sequence current and the processed cycle waves into a plurality of discriminators aiming at different fault causes to obtain classification probabilities of judging different types of fault causes, wherein the discriminators comprise a feature module and a classifier; Based on the classification probability of different types of fault causes, a final research and judgment result is selected according to actual requirements.
2. The method for analyzing fault causes of a small-current grounding system according to claim 1, wherein: The processing of the intercepted cycle wave specifically comprises the following steps: And carrying out resampling and normalization operation on the cut-out frequency, and carrying out FFT (fast Fourier transform) decomposition on the frequency subjected to resampling and normalization to obtain a zero-sequence voltage FFT result and a zero-sequence current FFT result.
3. The method for analyzing fault causes of a small-current grounding system according to claim 1, wherein: the discriminator comprises a feature module and a classifier, and the specific structure comprises: The characteristic module comprises an input layer, a characteristic extraction part, a characteristic fusion part and an output layer, wherein the input layer reads the zero sequence voltage, the zero sequence current and the processed cycle; The multiple discriminators share the network parameters of the same feature extraction part, and different discriminators have feature fusion parts and classifiers of different network parameters.
4. The method for analyzing fault cause of small current grounding system according to claim 3, wherein: The input format of the input layer is as follows: The format is [ m, n,4,256], wherein m represents the number of input samples, n represents the number of cycles of the number of input samples, 4 represents four channels, and 256 is the default number of nodes of one cycle; the four channels are sequentially and respectively input with zero sequence voltage, zero sequence voltage FFT result, zero sequence current and zero sequence current FFT result; The output format of the output layer is [ m,128], m represents the number of input samples, and 128 is the extracted abstract feature dimension.
5. The method for analyzing fault cause of small current grounding system according to claim 3, wherein: The feature extraction part has the structure that: The first part comprises a one-dimensional convolution layer, a batch normalization layer, reLu activation function layers and a one-dimensional maximum pooling layer, wherein the output of the first part is used as the input of the second part; the second part is a flattening layer, a linear transformation layer, a Dropout loss layer, reLu activation function layers.
6. The method for analyzing fault cause of small current grounding system according to claim 3, wherein: The specific structure of the characteristic fusion part is as follows: And carrying out element level multiplication on the output of the feature extraction part and the output of the feature extraction part through a linear transformation layer and a Softmax activation function layer, and obtaining the multiplication result through the linear transformation layer and a ReLu activation function layer to output.
7. The method for analyzing fault cause of small current grounding system according to claim 3, wherein: the specific structure of the classifier comprises: An input layer for receiving the output from the feature module, wherein the input format is [ m,128], m represents the number of input samples, and 128 is a feature dimension; The result of the input layer sequentially passes through a BatchNorm one-dimensional convolution layer, a linear transformation layer and a Sigmoid function layer to obtain the probability of belonging to the corresponding category.
8. A fault cause analysis system for a low-current grounding system by using the method of any one of claims 1-7, comprising a topology acquisition module, a device wave recording acquisition module, a wave recording processing module, a discriminator discriminating module and a studying and judging module, and being characterized in that: The topological structure acquisition module acquires the topological structure of the low-current ground fault system when the low-current ground fault occurs; The equipment wave recording acquisition module is used for finding out a fault point in the topological structure and acquiring equipment wave recording of equipment with the closest physical distance to the fault point; The wave recording processing module extracts the zero sequence voltage and zero sequence current in the wave recording of the equipment and the length of a single cycle; The discriminator discriminating module inputs the zero sequence voltage, the zero sequence current and the processed cycle waves into a plurality of discriminators aiming at different fault causes to obtain classifying probabilities of judging different types of fault causes, wherein the discriminators comprise a characteristic module and a classifier; And the research and judgment module is used for selecting a final research and judgment result according to actual requirements based on the classification probability of different types of fault causes.
9. A terminal comprises a processor and a storage medium, and is characterized in that: The storage medium is used for storing instructions; The processor being operative according to the instructions to perform the steps of the method according to any one of claims 1-7.
10. Computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the steps of the method according to any of claims 1-7.

Description

Fault cause analysis method and system for small-current grounding system Technical Field The invention belongs to the technical field of low-current grounding systems, and particularly relates to a method and a system for analyzing fault causes of a low-current grounding system. Background The low-current grounding system is a three-phase system in which a neutral point is not grounded, and the neutral point is grounded through an arc suppression coil or grounded with high impedance, and is also called as a neutral point indirect grounding system. When a phase fails to ground, the ground fault current tends to be much less than the load current, and such a system is referred to as a low current ground system, because a short circuit loop cannot be constructed. Common faults, such as single-phase grounding, wire breakage, short circuit and the like, are characterized by different characteristics and causes. Single-phase earth faults are represented by reduced fault phase voltage rather than increased fault phase voltage, and are usually caused by foreign matter bridging or tree obstacle, broken line faults cause abnormal voltage and unbalanced load, which are mostly caused by external force and lightning stroke, and short circuit faults are accompanied by current surge and protection tripping, and are mainly caused by insulation breakdown or foreign matter bridging. And the analysis of the cause after the fault occurrence can prevent similar faults from happening again and guide operation and maintenance personnel to take targeted measures (such as reinforcing insulation monitoring, optimizing lightning protection or fastening connection points), so that the reliability of the system is improved and the power failure loss is reduced. The fault in the low-current grounding system is deeply analyzed, and the cause analysis of the system is far from a routine program of post-responsibility, but is a strategic core link for improving the intelligent operation and maintenance level of the power grid and guaranteeing the power supply reliability. The prior art methods for analyzing and detecting faults occurring in a low current grounding system can be divided into two types, non-electrical quantity-based detection and electrical quantity-based detection. Based on non-electric quantity detection, the method mainly adopts non-electric quantity information for comprehensive research and judgment, for example Document 1 (rural distribution network tree line contradiction risk early warning and optimization processing considering strong convection weather, yao Fuxing, etc., and "technical and electrical science report") utilizes historical meteorological data and fault records to establish a prediction model to judge which areas are prone to faults in specific weather. Document 2 (electric power line inspection tree obstacle detection method design based on laser radar, li Junpeng, electronic design engineering) installs multispectral laser radar on equipment to collect the distance between tree and electric wire for judging whether tree obstacle grounding occurs. Based on electric quantity detection, the method mainly adopts information of electric quantity to build a database, and new faults are compared with existing data to obtain distances, wherein the closer the distances are, the higher the similarity of the secondary faults and the type or the secondary historical faults is. For example: document 3 (case-based failure diagnosis algorithm, kong Qin, etc., computer system application) proposes a case-based vector calculation based failure diagnosis algorithm based on case-based failure diagnosis algorithm, which performs one-to-one matching of failure information and case information by abstracting the failure information and performing similarity calculation with cases in a system case library. However, the prior art has some defects and shortcomings: 1. The hardware dependence on the non-electric quantity detection method is high, the cost is high, and the large-scale deployment is difficult, wherein the method is seriously dependent on external precise detection equipment such as a laser radar, satellite remote sensing, a high-definition camera, a meteorological sensor and the like. Not only are these devices expensive, they require a significant amount of capital investment for installation, calibration, maintenance, and subsequent data transmission and storage. This makes it difficult to achieve full coverage deployment in a power distribution network (especially a rural network or an old urban power network with a complex branch and structure), and a large number of monitoring dead zones exist. 2. The fault identification type based on the non-electric quantity detection method is limited, the sensing dimension is single, and the method is characterized in that external hidden danger is found through 'looking' and 'measuring'. The hidden faults such as 'gradual change' or 'hidden' faults which occur in eq