Multi-source data collaborative distribution network fault accurate perception system

Abstract

The present invention discloses a multi-source data collaborative distribution network fault accurate perception system, which relates to the field of smart grid technology and solves the technical problems that it is difficult to quickly locate the fault location and it is difficult to consider the impact of multiple factors on fault identification; by dividing the distribution network into grids, the power equipment and loads in each grid can be monitored more carefully, thereby improving the overall monitoring accuracy. By focusing on monitoring key nodes, the location of the fault can be preliminarily identified more quickly and the fault response time can be shortened. The collection of multi-source data provides rich information for fault analysis and helps to identify power grid faults. Multiple auxiliary monitoring nodes are set in each grid to assist the system in further fault location and increase the reliability of the system. By simultaneously collecting steady-state characteristic data, such as voltage, current, power, etc., as well as transient characteristic data such as instantaneous current waveform, harmonics, etc., it is convenient to fully understand the operating status of the power grid.

Description

Technical Field

[0001] The present invention belongs to the field of smart grids, and specifically relates to a multi-source data collaborative distribution network fault accurate perception system. Background Art

[0002] With the continuous growth of electricity demand and the widespread application of renewable energy, distribution networks are facing an increasingly complex operating environment. Distribution networks must not only meet users' demand for high reliability and high quality of electricity, but also need to deal with various potential faults and abnormal conditions. These faults may be caused by a variety of factors such as equipment aging, natural disasters, load fluctuations, etc., resulting in power outages or deterioration of power quality. Traditional distribution network monitoring systems often rely on a single data source (such as current, voltage, etc.), which makes it difficult to fully and accurately perceive fault conditions. With the development of information technology, emerging technologies such as the Internet of Things (IoT), big data analysis, and artificial intelligence have provided new opportunities for the monitoring and management of distribution networks. By integrating information from different sensors and data sources, real-time monitoring and intelligent analysis of the distribution network status can be achieved, thereby improving the accuracy and response speed of fault diagnosis.

[0003] Most power grid fault perception systems collect data at each node, and the scale of data collection is large. If the fault location is not at a node but between nodes, it is difficult to quickly locate the fault location, and it is also difficult to consider the impact of multiple factors on fault identification. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a multi-source data collaborative distribution network fault accurate perception system to solve the technical problems that it is difficult to quickly locate the fault position and it is difficult to consider the impact of multiple factors on fault identification.

[0005] To solve the above problems, the first aspect of the present invention provides a multi-source data collaborative distribution network fault accurate perception system, comprising:

[0006] Grid division module: divides the distribution network into several grids and obtains the grid topology of each grid;

[0007] Data collection module: selects key nodes and nodes in high-failure area in the grid as main monitoring nodes, randomly selects other nodes as auxiliary monitoring nodes, and collects multi-source data of the main monitoring nodes;

[0008] Among them, multi-source data collection is performed at the main monitoring node, including steady-state characteristic data and transient characteristic data, and at the same time, transient characteristic data collection is performed at the auxiliary monitoring node;

[0009] Grid fault diagnosis module: Generates an inter-grid fault diagnosis model based on the steady-state characteristic data and transient characteristic data detected by the main monitoring node, and identifies faults occurring in a grid or between grids based on the inter-grid fault diagnosis model;

[0010] Fault location module: The grid involved in the fault or the grid-to-grid fault is temporarily defined as the fault grid, and the grid fault in the fault grid is located based on the collected transient characteristic data of the grid or the grid-to-grid fault.

[0011] As a further solution of the present invention: the power grid division module divides the distribution network into a plurality of grids and obtains the power grid topology structure of each grid, including the following steps:

[0012] The distribution network is divided into several grids of the same size, and the total load of each grid is detected. If the total load of the grid exceeds a preset threshold, the grid is divided again until the total load of all grids is less than the preset threshold;

[0013] The boundaries of each grid are located to obtain the grid topology of the authorized grid.

[0014] As a further solution of the present invention: selecting key nodes and nodes in high-prone areas of faults in the grid as primary monitoring nodes, and randomly selecting other nodes as auxiliary monitoring nodes, including the following steps:

[0015] According to the grid topology, the degree and betweenness centrality of each node are counted, the degree and betweenness centrality of the node are standardized, the sum of the degree and betweenness centrality of the standardized node is calculated, and a list from high to low is generated, and the top 5% of the nodes in the list are selected as key nodes;

[0016] Count the failure frequencies of each node in the grid, and select the 5% nodes with the highest failure frequencies as the nodes in the high-failure area;

[0017] The key nodes and nodes in the high-prone-of-failure areas are selected as the main monitoring nodes, and other nodes are randomly selected as auxiliary monitoring nodes.

[0018] As a further solution of the present invention: collecting multi-source data at the main monitoring node, including steady-state characteristic data and transient characteristic data, and collecting transient characteristic data at the auxiliary monitoring node, including the following steps:

[0019] Collect multi-source data at the main monitoring node, including steady-state characteristic data and transient characteristic data;

[0020] Among them, the steady-state characteristics of the collected monitoring nodes include: node voltage, voltage amplitude and phase, node current, current amplitude and phase, active power, reactive power, apparent power and power factor;

[0021] A pulse signal of preset frequency and amplitude is introduced into the circuit of the acquisition monitoring node, and transient characteristic data of the pulse signal when it is reflected is detected, including voltage waveform, current waveform and power waveform;

[0022] By introducing a pulse signal of preset frequency and amplitude between auxiliary monitoring nodes, transient characteristic data of the pulse signal when reflected is detected.

[0023] As a further solution of the present invention: generating an inter-grid fault diagnosis model according to the steady-state characteristic data and transient characteristic data detected by the main monitoring node, and identifying the faults occurring in the grid or between grids according to the inter-grid fault diagnosis model, including the following steps:

[0024] Detect voltage waveform, current waveform and power waveform in transient characteristic data between all main monitoring nodes, and extract the harmonic content of each waveform;

[0025] Obtain the historical data of steady-state characteristic data and transient characteristic data of node detection, extract the harmonic content of voltage waveform, current waveform and power waveform in the transient characteristic data, and add labels of output fault data and output normal data to the historical data of fault occurrence and non-fault occurrence;

[0026] Build an inter-grid fault diagnosis model by training a machine learning model with labeled historical data;

[0027] The steady-state characteristic data collected in real time based on the detection of the main monitoring nodes and the harmonic content data extracted from the transient characteristic data are input into the constructed inter-grid fault diagnosis model to identify whether there is a fault between the corresponding main monitoring nodes;

[0028] If the fault between the main monitoring nodes occurs in the same grid, the fault in the grid is determined;

[0029] If the faults between the main monitoring nodes occur in different grids, the faults occurring between the corresponding grids are determined.

[0030] As a further solution of the present invention: detecting voltage waveforms, current waveforms and power waveforms in transient characteristic data between all main monitoring nodes, and extracting the harmonic content of each waveform, including the following steps:

[0031] Detecting voltage waveforms, current waveforms and power waveforms in transient characteristic data between all main monitoring nodes, and dividing the waveform time domain signals into a number of windows, each window containing a number of cycles of signals;

[0032] The time domain signal in each window is converted into a frequency domain signal through FFT fast Fourier transform;

[0033] The amplitude spectrum and phase spectrum are extracted from the frequency domain signal after FFT fast Fourier transform; the amplitude spectrum shows the intensity of each frequency component, while the phase spectrum shows their phase information.

[0034] Mark the fundamental wave and its harmonic parts through the amplitude spectrum and phase spectrum;

[0035] Harmonics are usually expressed as integer multiples of the fundamental frequency. The harmonics marked here include: 2nd harmonic, 3rd harmonic and above.

[0036] The harmonic content of each waveform is extracted using the following formula:

[0037]

[0038] Where f is the harmonic content of each waveform, Vi is the effective value of the i-th harmonic, V0 is the effective value of the fundamental wave, i∈(1,2,…,n), and n is the total number of marked harmonics.

[0039] As a further solution of the present invention, a method of identifying a fault occurring in a grid or between grids according to an inter-grid fault diagnosis model, temporarily defining a grid involved in a fault occurring in a grid or between grids as a faulty grid, and locating a power grid fault in the faulty grid according to the collected transient characteristic data of the grid or between grids having the fault, comprises the following steps:

[0040] Identify the faults occurring in the grid or between grids according to the inter-grid fault diagnosis model, and temporarily define the grid involved in the faults occurring in the grid or between grids as the faulty grid;

[0041] According to the auxiliary monitoring nodes of the grid or between grids, the auxiliary monitoring nodes perform transient characteristic data detection in pairs;

[0042] Analyze the inter-node fault assessment coefficient based on the collected transient characteristic data between auxiliary monitoring nodes;

[0043] If the inter-node fault assessment coefficient between the auxiliary monitoring nodes is greater than a preset threshold, it is determined that there is a fault between the auxiliary monitoring nodes, otherwise it is determined that there is no fault between the auxiliary monitoring nodes;

[0044] For auxiliary monitoring nodes with faults, if the two auxiliary monitoring nodes belong to different grids, the auxiliary monitoring node in the grid with the highest fault frequency within the preset time period is determined as the base detection node; if the two auxiliary monitoring nodes belong to the same grid, the one with the highest total load among the two auxiliary monitoring nodes is selected as the base detection node, and the other node of the two auxiliary monitoring nodes is selected as the moving point detection node;

[0045] From the initial moving point detection node to the direction close to the base point detection node, by traversing each node between the initial moving point detection node and the base point detection node, the nodes between the initial moving point detection node and the base point detection node are selected in turn as new moving point detection nodes, and the inter-node fault assessment coefficients of the base point detection node and the moving point detection node are analyzed;

[0046] If the inter-node fault assessment coefficient obtained during the movement of the moving point detection node drops by more than a preset threshold, the power grid fault is located between two moving point detection nodes whose drop exceeds the preset threshold;

[0047] If the fluctuation of the inter-node fault assessment coefficient obtained during the movement of the moving point detection node does not drop below a preset threshold, the power grid fault is located between the last detected moving point detection node and the base point detection node.

[0048] As a further solution of the present invention: analyzing the inter-node fault assessment coefficient according to the collected transient characteristic data between the auxiliary monitoring nodes, including the following steps:

[0049] Extract the harmonic content of voltage waveform, current waveform and power waveform in transient characteristic data;

[0050] The transient characteristic data detection time between the auxiliary monitoring nodes is divided into several detection time intervals. According to the voltage waveform, current waveform and power waveform in the transient characteristic data, the effective value and peak value of the current waveform in each time interval are extracted respectively, the effective value, peak value and phase difference between the voltage and current of the voltage waveform in each time interval are extracted, the mean value and peak value of the power waveform in each time interval are extracted, and the harmonic content corresponding to the voltage waveform, current waveform and power waveform are extracted;

[0051] The inter-node fault assessment coefficient is analyzed using the following formula:

[0052]

[0053] Among them, Q is the fault assessment coefficient between nodes, Si1 is the variance of the effective value of the current waveform in each time interval, μi1 is the mean of the effective value of the current waveform in each time interval, Sv1 is the variance of the effective value of the voltage waveform in each time interval, μv1 is the mean of the effective value of the voltage waveform in each time interval, Sp1 is the variance of the mean of the power waveform in each time interval, μp1 is the average of the mean of the power waveform in each time interval, Si2 is the variance of the peak value of the current waveform in each time interval, μi2 is the mean of the peak value of the current waveform in each time interval, Sv2 is the variance of the peak value of the voltage waveform in each time interval, μv2 is the mean of the peak value of the voltage waveform in each time interval, Sp2 is the variance of the peak value of the power waveform in each time interval, μp2 is the average of the peak value of the power waveform in each time interval, θv is the phase angle of the voltage waveform, θi is the phase angle of the current waveform, and f0 is the mean of the harmonic content corresponding to the voltage waveform, current waveform and power waveform.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] By dividing the distribution network into grids, the present invention can perform more detailed monitoring of the power equipment and loads in each grid, thereby improving the overall monitoring accuracy. By focusing on monitoring key nodes, the location of the fault can be preliminarily identified more quickly, shortening the fault response time. The collection of multi-source data provides rich information for fault analysis and helps to identify power grid faults. Multiple auxiliary monitoring nodes are set in each grid to assist the system in further fault location and increase the reliability of the system.

[0056] The present invention facilitates a comprehensive understanding of the operating status of the power grid by simultaneously collecting steady-state characteristic data, such as voltage, current, power, etc., and transient characteristic data, such as instantaneous current waveform, harmonics, etc. This comprehensive data collection helps to identify faults more accurately. Steady-state characteristic data provides the basic situation of the long-term operation of the system, while transient characteristic data reflects the response of the system under instantaneous changes. Combining the data of the two, the cause of the fault can be analyzed more comprehensively, thereby improving the accuracy of fault diagnosis. Through the inter-grid fault diagnosis model, it is convenient to quickly locate the node of the specific fault, thereby reducing the power outage time and improving the reliability of power supply. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 It is a schematic diagram of the system framework of the present invention;

[0059] Figure 2 A flow chart of the method for setting the main monitoring node and the auxiliary monitoring node of the present invention. DETAILED DESCRIPTION

[0060] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] See also Figure 1-Figure 2 The first embodiment of the present invention provides a multi-source data collaborative distribution network fault accurate perception system, including:

[0062] Grid division module: divides the distribution network into several grids and obtains the grid topology of each grid;

[0063] Data collection module: selects key nodes and nodes in high-failure area in the grid as main monitoring nodes, randomly selects other nodes as auxiliary monitoring nodes, and collects multi-source data of the main monitoring nodes;

[0064] Among them, multi-source data collection is performed at the main monitoring node, including steady-state characteristic data and transient characteristic data, and at the same time, transient characteristic data collection is performed at the auxiliary monitoring node;

[0065] Grid fault diagnosis module: Generates an inter-grid fault diagnosis model based on the steady-state characteristic data and transient characteristic data detected by the main monitoring node, and identifies faults occurring in a grid or between grids based on the inter-grid fault diagnosis model;

[0066] Fault location module: The grid involved in the fault or the grid-to-grid fault is temporarily defined as the fault grid, and the grid fault in the fault grid is located based on the collected transient characteristic data of the grid or the grid-to-grid fault.

[0067] Specifically, in this embodiment, the grids are divided according to the load density, and the size of each grid is determined.

[0068] Define key nodes and other high-failure areas in each grid to set up monitoring nodes. Use historical failure records to determine nodes with higher failure rates through statistical analysis and identify nodes in areas with frequent failures.

[0069] Collect the topological data of the distribution network, including the connection relationship between each node, cable type and capacity, etc. Use graph theory to construct the topological graph of each grid, where nodes represent devices and edges represent connections.

[0070] The main monitoring nodes are determined from key nodes and nodes in high-failure areas to ensure that these nodes can effectively reflect the operating status of the entire grid.

[0071] Other nodes are randomly selected as auxiliary monitoring nodes to supplement the data collection of the main monitoring.

[0072] Various sensors, such as current sensors, voltage sensors, etc., are installed at the main monitoring and auxiliary monitoring nodes to obtain real-time data.

[0073] The power distribution network is divided into several grids to obtain the power grid topology of each grid; key nodes and nodes in high-prone fault areas are selected as main monitoring nodes in the grid, and other nodes are randomly selected as auxiliary monitoring nodes to collect multi-source data of the main monitoring nodes;

[0074] By dividing the distribution network into grids, the power equipment and loads within each grid can be monitored in more detail, thereby improving the overall monitoring accuracy.

[0075] By focusing on monitoring key nodes, the location of the fault can be identified more quickly and the fault response time can be shortened. The collection of multi-source data provides rich information for fault analysis and helps to identify power grid faults. Multiple auxiliary monitoring nodes are set up in each grid to assist the system in further fault location and increase system reliability.

[0076] Collect multi-source data at the main monitoring node, including steady-state characteristic data and transient characteristic data, and collect transient characteristic data at the auxiliary monitoring node; generate an inter-grid fault diagnosis model based on the steady-state characteristic data and transient characteristic data detected by the main monitoring node, and identify faults occurring in a grid or between grids based on the inter-grid fault diagnosis model;

[0077] By simultaneously collecting steady-state characteristic data, such as voltage, current, power, etc., as well as transient characteristic data, such as instantaneous current waveform, harmonics, etc., it is convenient to fully understand the operating status of the power grid. This comprehensive data collection helps to identify faults more accurately.

[0078] Steady-state characteristic data provides the basic situation of the long-term operation of the system, while transient characteristic data reflects the response of the system under instantaneous changes. Combining the data of the two can more comprehensively analyze the cause of the fault, thereby improving the accuracy of fault diagnosis.

[0079] By analyzing different types of data and using machine learning and pattern recognition technology, a more accurate fault diagnosis model can be built.

[0080] The inter-grid fault diagnosis model makes it easy to quickly locate the specific fault node, thereby reducing power outage time and improving power supply reliability.

[0081] In one embodiment of the present invention, a distribution network is divided into a plurality of grids, and a grid topology structure of each grid is obtained, including the following steps:

[0082] The distribution network is divided into several grids of the same size, and the total load of each grid is detected. If the total load of the grid exceeds a preset threshold, the grid is divided again until the total load of all grids is less than the preset threshold;

[0083] The boundaries of each grid are located to obtain the grid topology of the authorized grid.

[0084] In one embodiment of the present invention, key nodes and nodes in high-prone-to-failure areas are selected in a grid as primary monitoring nodes, and other nodes are randomly selected as auxiliary monitoring nodes, including the following steps:

[0085] According to the grid topology, the degree and betweenness centrality of each node are counted, the degree and betweenness centrality of the node are standardized, the sum of the degree and betweenness centrality of the standardized node is calculated, and a list from high to low is generated, and the top 5% of the nodes in the list are selected as key nodes;

[0086] Count the failure frequencies of each node in the grid, and select the 5% nodes with the highest failure frequencies as the nodes in the high-failure area;

[0087] The key nodes and nodes in the high-prone-of-failure areas are selected as the main monitoring nodes, and other nodes are randomly selected as auxiliary monitoring nodes.

[0088] Specifically, in this embodiment, the degree Cd1 is the number of edges directly connected to a node, and the betweenness centrality Cb1 is the number of shortest paths passing through the node. The degree and betweenness centrality of the node are standardized using the following formula:

[0089] Cd2=(Cd1-Cdmin) / (Cdmax-Cdmin);

[0090] Cb2=(Cb1-Cbmin) / (Cbmax-Cbmin);

[0091] Among them, Cd2 is the standardized degree, Cb2 is the standardized betweenness centrality, Cdmin and Cdmax are the minimum and maximum values ​​of the degree of all nodes, respectively, and Cbmin and Cbmax are the minimum and maximum values ​​of the betweenness centrality of all nodes, respectively.

[0092] In one embodiment of the present invention, multi-source data collection including steady-state characteristic data and transient characteristic data is performed at the main monitoring node, and transient characteristic data collection is performed at the auxiliary monitoring node, including the following steps:

[0093] Collect multi-source data at the main monitoring node, including steady-state characteristic data and transient characteristic data;

[0094] Among them, the steady-state characteristics of the collected monitoring nodes include: node voltage, voltage amplitude and phase, node current, current amplitude and phase, active power, reactive power, apparent power and power factor;

[0095] A pulse signal of preset frequency and amplitude is introduced into the circuit of the acquisition monitoring node, and transient characteristic data of the pulse signal when it is reflected is detected, including voltage waveform, current waveform and power waveform;

[0096] By introducing a pulse signal of preset frequency and amplitude between auxiliary monitoring nodes, transient characteristic data of the pulse signal when reflected is detected;

[0097] The preset frequency of the pulse signal should be distinguished from the power frequency and its harmonic components. Usually, a high-frequency pulse signal is selected to reduce the impact of power frequency noise on detection. For instantaneous faults, such as lightning strikes, a high-frequency pulse greater than 10 kHz is selected to capture fast transient characteristics. For continuous faults, such as line breaks or grounding faults, a medium or low frequency of 1kHz to 10 kHz is selected to take into account both propagation distance and resolution.

[0098] In one embodiment of the present invention, a grid-to-grid fault diagnosis model is generated according to steady-state characteristic data and transient characteristic data detected by a main monitoring node, and a grid-to-grid fault diagnosis model is used to identify a fault occurring in a grid or between grids, including the following steps:

[0099] Detect voltage waveform, current waveform and power waveform in transient characteristic data between all main monitoring nodes, and extract the harmonic content of each waveform;

[0100] Obtain the historical data of steady-state characteristic data and transient characteristic data of node detection, extract the harmonic content of voltage waveform, current waveform and power waveform in the transient characteristic data, and add labels of output fault data and output normal data to the historical data of fault occurrence and non-fault occurrence;

[0101] Build an inter-grid fault diagnosis model by training a machine learning model with labeled historical data;

[0102] The steady-state characteristic data collected in real time based on the detection of the main monitoring nodes and the harmonic content data extracted from the transient characteristic data are input into the constructed inter-grid fault diagnosis model to identify whether there is a fault between the corresponding main monitoring nodes;

[0103] If the fault between the main monitoring nodes occurs in the same grid, the fault in the grid is determined;

[0104] If the faults between the main monitoring nodes occur in different grids, the faults occurring between the corresponding grids are determined.

[0105] Specifically, in this embodiment, the transient response of the system to the pulse signal is recorded by a high-precision data acquisition device, and sampling is continued within a certain time detection interval. The recorded data should include voltage, current waveforms and power waveforms, and the collected signals are denoised to eliminate external interference, so as to improve the accuracy of subsequent analysis.

[0106] In one embodiment of the present invention, detecting voltage waveforms, current waveforms and power waveforms in transient characteristic data between all main monitoring nodes and extracting harmonic content of each waveform includes the following steps:

[0107] Detecting voltage waveforms, current waveforms and power waveforms in transient characteristic data between all main monitoring nodes, and dividing the waveform time domain signals into a number of windows, each window containing a number of cycles of signals;

[0108] The time domain signal in each window is converted into a frequency domain signal through FFT fast Fourier transform;

[0109] The amplitude spectrum and phase spectrum are extracted from the frequency domain signal after the FFT fast Fourier transform; specifically, in this embodiment, the amplitude spectrum displays the intensity of each frequency component, while the phase spectrum displays their phase information.

[0110] Mark the fundamental wave and its harmonic parts through the amplitude spectrum and phase spectrum;

[0111] In this embodiment, harmonics are usually expressed as integer multiples of the fundamental frequency, and the harmonics marked here include: 2nd harmonics, 3rd harmonics and above.

[0112] The harmonic content of each waveform is extracted using the following formula:

[0113]

[0114] Where f is the harmonic content of each waveform, Vi is the effective value of the i-th harmonic, V0 is the effective value of the fundamental wave, i∈(1,2,…,n), and n is the total number of marked harmonics.

[0115] In one embodiment of the present invention, a fault occurring in a grid or between grids is identified according to an inter-grid fault diagnosis model, a grid involved in a fault occurring in a grid or between grids is temporarily determined as a faulty grid, and a power grid fault is located in the faulty grid according to the collected transient characteristic data of the grid or between grids where the fault occurs, including the following steps:

[0116] Identify the faults occurring in the grid or between grids according to the inter-grid fault diagnosis model, and temporarily define the grid involved in the faults occurring in the grid or between grids as the faulty grid;

[0117] According to the auxiliary monitoring nodes of the grid or between grids, the auxiliary monitoring nodes perform transient characteristic data detection in pairs;

[0118] Analyze the inter-node fault assessment coefficient based on the collected transient characteristic data between auxiliary monitoring nodes;

[0119] If the inter-node fault assessment coefficient between the auxiliary monitoring nodes is greater than a preset threshold, it is determined that there is a fault between the auxiliary monitoring nodes, otherwise it is determined that there is no fault between the auxiliary monitoring nodes;

[0120] In this embodiment, the threshold of the inter-node fault assessment coefficient is determined to be 0.5 by statistically analyzing the inter-node fault assessment coefficients of all faulty and normal nodes in historical data. If the inter-node fault assessment coefficient between the auxiliary monitoring nodes is greater than 0.5, it is determined that there is a fault between the auxiliary monitoring nodes; otherwise, it is determined that there is no fault between the auxiliary monitoring nodes.

[0121] For auxiliary monitoring nodes with faults, if the two auxiliary monitoring nodes belong to different grids, the auxiliary monitoring node in the grid with the highest fault frequency within the preset time period is determined as the base detection node; if the two auxiliary monitoring nodes belong to the same grid, the one with the highest total load among the two auxiliary monitoring nodes is selected as the base detection node, and the other node of the two auxiliary monitoring nodes is selected as the moving point detection node;

[0122] From the initial moving point detection node to the direction close to the base point detection node, by traversing each node between the initial moving point detection node and the base point detection node, the nodes between the initial moving point detection node and the base point detection node are selected in turn as new moving point detection nodes, and the inter-node fault assessment coefficients of the base point detection node and the moving point detection node are analyzed;

[0123] If the inter-node fault assessment coefficient obtained during the movement of the moving point detection node drops by more than a preset threshold, the power grid fault is located between two moving point detection nodes whose drop exceeds the preset threshold;

[0124] If the fluctuation of the inter-node fault assessment coefficient obtained during the movement of the moving point detection node does not drop below a preset threshold, the power grid fault is located between the last detected moving point detection node and the base point detection node.

[0125] In this embodiment, the inter-node fault assessment coefficient fluctuates by a preset threshold value, which is determined to be 0.2 by counting the inter-node fault assessment coefficients of all faulty and normal nodes in historical data. If the inter-node fault assessment coefficient obtained during the movement of the moving point detection node fluctuates by more than 0.2, the power grid fault is located between two moving point detection nodes whose fluctuation exceeds the preset threshold value;

[0126] If the inter-node fault assessment coefficient fluctuation obtained during the movement of the moving point detection node does not exceed 0.2, the power grid fault is located between the last detected moving point detection node and the base point detection node.

[0127] In one embodiment of the present invention, analyzing the inter-node fault assessment coefficient according to the collected transient characteristic data between the auxiliary monitoring nodes includes the following steps:

[0128] Extract the harmonic content of voltage waveform, current waveform and power waveform in transient characteristic data;

[0129] The transient characteristic data detection time between the auxiliary monitoring nodes is divided into several detection time intervals. According to the voltage waveform, current waveform and power waveform in the transient characteristic data, the effective value and peak value of the current waveform in each time interval are extracted respectively, the effective value, peak value and phase difference between the voltage and current of the voltage waveform in each time interval are extracted, the mean value and peak value of the power waveform in each time interval are extracted, and the harmonic content corresponding to the voltage waveform, current waveform and power waveform are extracted;

[0130] The inter-node fault assessment coefficient is analyzed using the following formula:

[0131]

[0132] Among them, Q is the fault assessment coefficient between nodes, Si1 is the variance of the effective value of the current waveform in each time interval, μi1 is the mean of the effective value of the current waveform in each time interval, Sv1 is the variance of the effective value of the voltage waveform in each time interval, μv1 is the mean of the effective value of the voltage waveform in each time interval, Sp1 is the variance of the mean of the power waveform in each time interval, μp1 is the average of the mean of the power waveform in each time interval, Si2 is the variance of the peak value of the current waveform in each time interval, μi2 is the mean of the peak value of the current waveform in each time interval, Sv2 is the variance of the peak value of the voltage waveform in each time interval, μv2 is the mean of the peak value of the voltage waveform in each time interval, Sp2 is the variance of the peak value of the power waveform in each time interval, μp2 is the average of the peak value of the power waveform in each time interval, θv is the phase angle of the voltage waveform, θi is the phase angle of the current waveform, and f0 is the mean of the harmonic content corresponding to the voltage waveform, current waveform and power waveform.

[0133] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A multi-source data collaborative distribution network fault accurate perception system, characterized by: include: Grid division module: divides the distribution network into several grids and obtains the grid topology of each grid; Data collection module: selects key nodes and nodes in high-failure area in the grid as main monitoring nodes, randomly selects other nodes as auxiliary monitoring nodes, and collects multi-source data of the main monitoring nodes; Among them, according to the historical data of node failure frequency in the grid, the nodes in the high-incidence area of ​​failure are screened, and according to the degree and betweenness centrality of the nodes, the key nodes are screened, and the screened nodes are used as the main monitoring nodes; multi-source data are collected at the main monitoring nodes, including steady-state characteristic data and transient characteristic data, and at the same time, transient characteristic data are collected at the auxiliary monitoring nodes; Grid fault diagnosis module: Generates an inter-grid fault diagnosis model based on the steady-state characteristic data and transient characteristic data detected by the main monitoring node, and identifies faults occurring in a grid or between grids based on the inter-grid fault diagnosis model; wherein, the inter-grid fault diagnosis model is constructed based on the harmonic content of the voltage waveform, current waveform, and power waveform in the steady-state characteristic data and transient characteristic data; Fault location module: The grid involved in the fault in the grid or between grids is temporarily defined as the fault grid. According to the transient characteristic data collected between the grid where the fault occurs or the auxiliary monitoring nodes between grids, the base point detection node is selected according to the fault frequency of the node, and the dynamic point detection node is selected in the grid or the auxiliary monitoring nodes between grids. According to the detection data between nodes, the fault between nodes is evaluated, and the power grid fault in the fault grid is located; The grid division module divides the distribution network into several grids and obtains the grid topology structure of each grid, including the following steps: The distribution network is divided into several grids of the same size, and the total load of each grid is detected. If the total load of the grid exceeds a preset threshold, the grid is divided again until the total load of all grids is less than the preset threshold; Locate the boundaries of each grid and obtain the grid topology of the authorized grid; In the grid, key nodes and nodes in high-prone areas of faults are selected as primary monitoring nodes, and other nodes are randomly selected as auxiliary monitoring nodes, including the following steps: According to the grid topology, the degree and betweenness centrality of each node are counted, the degree and betweenness centrality of the node are standardized, the sum of the degree and betweenness centrality of the standardized node is calculated, and a list from high to low is generated, and the top 5% of the nodes in the list are selected as key nodes; Count the failure frequencies of each node in the grid, and select the 5% nodes with the highest failure frequencies as the nodes in the high-failure area; Use key nodes and nodes in high-failure area as primary monitoring nodes, and randomly select other nodes as auxiliary monitoring nodes; According to the inter-grid fault diagnosis model, the fault occurring in the grid or between grids is identified, the grid involved in the fault occurring in the grid or between grids is temporarily determined as the faulty grid, and the power grid fault in the faulty grid is located according to the collected transient characteristic data of the grid or between grids where the fault occurs, including the following steps: Identify the faults occurring in the grid or between grids according to the inter-grid fault diagnosis model, and temporarily define the grid involved in the faults occurring in the grid or between grids as the faulty grid; According to the auxiliary monitoring nodes of the grid or between grids, the auxiliary monitoring nodes perform transient characteristic data detection in pairs; Analyze the inter-node fault assessment coefficient based on the collected transient characteristic data between auxiliary monitoring nodes; If the inter-node fault assessment coefficient between the auxiliary monitoring nodes is greater than a preset threshold, it is determined that there is a fault between the auxiliary monitoring nodes, otherwise it is determined that there is no fault between the auxiliary monitoring nodes; For auxiliary monitoring nodes with faults, if the two auxiliary monitoring nodes belong to different grids, the auxiliary monitoring node in the grid with the highest fault frequency within the preset time period is determined as the base detection node; if the two auxiliary monitoring nodes belong to the same grid, the one with the highest total load among the two auxiliary monitoring nodes is selected as the base detection node, and the other node of the two auxiliary monitoring nodes is selected as the moving point detection node; From the initial moving point detection node to the direction close to the base point detection node, by traversing each node between the initial moving point detection node and the base point detection node, the nodes between the initial moving point detection node and the base point detection node are selected in turn as new moving point detection nodes, and the inter-node fault assessment coefficients of the base point detection node and the moving point detection node are analyzed; If the inter-node fault assessment coefficient obtained during the movement of the moving point detection node drops by more than a preset threshold, the power grid fault is located between two moving point detection nodes whose drop exceeds the preset threshold; If the fluctuation of the inter-node fault assessment coefficient obtained during the movement of the moving point detection node does not drop below a preset threshold, the power grid fault is located between the last detected moving point detection node and the base point detection node.

2. The multi-source data collaborative distribution network fault accurate perception system according to claim 1 is characterized in that: Collecting multi-source data at the main monitoring node, including steady-state characteristic data and transient characteristic data, and collecting transient characteristic data at the auxiliary monitoring node, includes the following steps: Collect multi-source data at the main monitoring node, including steady-state characteristic data and transient characteristic data; Among them, the steady-state characteristics of the collected monitoring nodes include: node voltage, voltage amplitude and phase, node current, current amplitude and phase, active power, reactive power, apparent power and power factor; A pulse signal of preset frequency and amplitude is introduced into the circuit of the acquisition monitoring node, and transient characteristic data of the pulse signal when it is reflected is detected, including voltage waveform, current waveform and power waveform; By introducing a pulse signal of preset frequency and amplitude between auxiliary monitoring nodes, transient characteristic data of the pulse signal when reflected is detected; Among them, the pulse signal of preset frequency and amplitude, the preset frequency is set to be different from the power frequency and its harmonic components and greater than 1kHz, and the preset amplitude setting range is: low voltage distribution network is 100V-1kV, and medium and high voltage distribution network is 1kV-10kV.

3. The multi-source data collaborative distribution network fault accurate perception system according to claim 2 is characterized in that: According to the steady-state characteristic data and transient characteristic data detected by the main monitoring node, an inter-grid fault diagnosis model is generated, and the fault occurring in the grid or between grids is identified according to the inter-grid fault diagnosis model, including the following steps: Detect voltage waveform, current waveform and power waveform in transient characteristic data between all main monitoring nodes, and extract the harmonic content of each waveform; Obtain the historical data of steady-state characteristic data and transient characteristic data of node detection, extract the harmonic content of voltage waveform, current waveform and power waveform in the transient characteristic data, and add labels of output fault data and output normal data to the historical data of fault occurrence and non-fault occurrence; Build an inter-grid fault diagnosis model by training a machine learning model with labeled historical data; The steady-state characteristic data collected in real time based on the detection of the main monitoring nodes and the harmonic content data extracted from the transient characteristic data are input into the constructed inter-grid fault diagnosis model to identify whether there is a fault between the corresponding main monitoring nodes; If the fault between the main monitoring nodes occurs in the same grid, the fault in the grid is determined; If the faults between the main monitoring nodes occur in different grids, the faults occurring between the corresponding grids are determined.

4. The multi-source data collaborative distribution network fault accurate perception system according to claim 3 is characterized in that: Detecting voltage waveforms, current waveforms, and power waveforms in transient characteristic data between all main monitoring nodes and extracting the harmonic content of each waveform includes the following steps: Detecting voltage waveforms, current waveforms and power waveforms in transient characteristic data between all main monitoring nodes, and dividing the waveform time domain signals into a number of windows, each window containing a number of cycles of signals; The time domain signal in each window is converted into a frequency domain signal through FFT fast Fourier transform; Extract the amplitude spectrum and phase spectrum from the frequency domain signal after FFT fast Fourier transform; Mark the fundamental wave and its harmonic parts through the amplitude spectrum and phase spectrum; The harmonic content of each waveform is extracted using the following formula: ； Where f is the harmonic content of each waveform, Vi is the effective value of the i-th harmonic, V0 is the effective value of the fundamental wave, i∈(1,2,…,n), and n is the total number of marked harmonics.

5. The multi-source data collaborative distribution network fault accurate perception system according to claim 4 is characterized in that: According to the collected transient characteristic data between the auxiliary monitoring nodes, the inter-node fault assessment coefficient is analyzed, including the following steps: Extract the harmonic content of voltage waveform, current waveform and power waveform in transient characteristic data; The transient characteristic data detection time between the auxiliary monitoring nodes is divided into several detection time intervals. According to the voltage waveform, current waveform and power waveform in the transient characteristic data, the effective value and peak value of the current waveform in each time interval are extracted respectively, the effective value, peak value and phase difference between the voltage and current of the voltage waveform in each time interval are extracted, the mean value and peak value of the power waveform in each time interval are extracted, and the harmonic content corresponding to the voltage waveform, current waveform and power waveform are extracted; The inter-node fault assessment coefficient is analyzed using the following formula: ； Among them, Q is the fault assessment coefficient between nodes, Si1 is the variance of the effective value of the current waveform in each time interval, μi1 is the mean of the effective value of the current waveform in each time interval, Sv1 is the variance of the effective value of the voltage waveform in each time interval, μv1 is the mean of the effective value of the voltage waveform in each time interval, Sp1 is the variance of the mean of the power waveform in each time interval, μp1 is the average of the mean of the power waveform in each time interval, Si2 is the variance of the peak value of the current waveform in each time interval, μi2 is the mean of the peak value of the current waveform in each time interval, Sv2 is the variance of the peak value of the voltage waveform in each time interval, μv2 is the mean of the peak value of the voltage waveform in each time interval, Sp2 is the variance of the peak value of the power waveform in each time interval, μp2 is the average of the peak value of the power waveform in each time interval, θv is the phase angle of the voltage waveform, θi is the phase angle of the current waveform, and f0 is the mean of the harmonic content corresponding to the voltage waveform, current waveform and power waveform.

Images