GIS partial discharge signal detection method, device and storage medium

Abstract

The present invention discloses a GIS local discharge signal detection method, device and storage medium. The method comprises the following steps: collecting electromagnetic field signals inside the GIS, using the time-frequency domain features of the electromagnetic field signals as nodes of a graph neural network for correlation strength screening, and obtaining time-domain strong correlation features and frequency-domain strong correlation features, wherein the lines between the nodes represent the relationship between the features represented by the nodes; extracting the time-domain strong correlation features of the electromagnetic field signals, and determining that a local discharge signal is monitored based on the time-domain strong correlation features, and outputting a first time-domain waveform diagram during discharge; extracting the frequency-domain strong correlation features of the electromagnetic field signals, and determining that a local discharge signal is monitored based on the frequency-domain strong correlation features, and outputting a frequency-domain waveform diagram during discharge; obtaining a global-domain waveform diagram of the local discharge signal based on the first time-domain waveform diagram and the frequency-domain waveform diagram; the present invention can improve the accuracy of local discharge signal detection while reducing the amount of calculation.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment, and in particular to a GIS partial discharge signal detection method, device and storage medium. Background Art

[0002] In high-voltage gas insulated switchgear (GIS), partial discharge is a common fault phenomenon that can cause equipment performance degradation, damage, and even accidents. Accurate and rapid collection of partial discharge signals inside GIS is crucial for equipment health status monitoring and fault diagnosis.

[0003] Most of the existing partial discharge signal acquisition devices use PRPD (Phase Resolved Partial Discharge) discharge spectra and other forms to transmit compressed partial discharge signals to the host computer, that is, the frame where the discharge data is located is subjected to Fast Fourier Transform (FFT) to extract the spectrum, and the obtained discharge frequency assignment and phase data are sent to the host computer, which cannot restore the detected high-frequency, complete full-time waveform partial discharge point time domain signal to the host computer, and cannot meet the needs of accurate positioning of partial discharge signals inside GIS. Since the sampling rate is proportional to the positioning accuracy, improving the sampling accuracy will face problems such as insufficient chip computing resources. Therefore, due to the defects of low sampling frequency, weak detection performance, and huge data volume, the existing GIS partial discharge monitoring on the market has problems such as missed reports and high false alarm rates.

[0004] In the related art, the patent application with publication number CN103558519A proposed a GIS partial discharge ultrasonic signal recognition method, which pre-processed the known samples of GIS partial discharge ultrasonic signal, then extracted the discharge characteristic parameters such as average amplitude, root mean square, peak index, kurtosis, waveform index, pulse index, margin index, and finally established a fuzzy logic clustering neural network to pre-process the GIS partial discharge ultrasonic signal to be identified. However, this scheme uses a neural network to monitor UHF signals instead of screening the time-frequency domain indicators of UHF signals. Although the patent application document with publication number CN103558519A proposed to use "average amplitude, root mean square, peak index, kurtosis, waveform index, pulse index, margin index" to monitor the ultrasonic discharge waveform, it did not give the basis for selecting the indicators. Summary of the invention

[0005] The technical problem to be solved by the present invention is how to improve the accuracy of partial discharge signal detection while reducing the amount of calculation.

[0006] The present invention solves the above technical problems by the following technical means:

[0007] The present invention proposes a GIS partial discharge signal detection method, the method comprising:

[0008] The electromagnetic field signals inside the GIS are collected, and the time-frequency domain features of the electromagnetic field signals are used as nodes of the graph neural network to screen the correlation strength, thereby obtaining strong correlation features in the time domain and strong correlation features in the frequency domain, wherein the lines between the nodes represent the relationship between the features represented by the nodes;

[0009] Extracting the time-domain strong correlation feature from the electromagnetic field signal, determining that a partial discharge signal is monitored based on the time-domain strong correlation feature, and outputting a first time-domain waveform diagram during discharge;

[0010] Extracting the frequency domain strong correlation feature from the electromagnetic field signal, determining that a partial discharge signal is monitored based on the frequency domain strong correlation feature, and outputting a frequency domain waveform diagram during discharge;

[0011] Based on the first time domain waveform diagram and the frequency domain waveform diagram, a full domain waveform diagram of the partial discharge signal is obtained.

[0012] Furthermore, the time-frequency domain features include time domain features and frequency domain features. The time domain features include margin, skewness, root mean square, mean, variance, peak-to-peak value, shape factor, kurtosis, impulse factor and C index; the frequency domain features include mean frequency, frequency concentration, frequency center, root mean square frequency and main frequency band position change.

[0013] Furthermore, the time-frequency domain features of the electromagnetic field signal are used as nodes of the graph neural network to perform correlation strength screening to obtain time-domain strong correlation features and frequency-domain strong correlation features, including:

[0014] Complete feature map in graph neural network middle, is a node set, is the edge set, let Respectively represent nodes The eigenvectors of the edges, the eigenvectors of the nodes and the state vectors of its surrounding nodes and nodes The feature vectors of the surrounding nodes;

[0015] According to the local transfer function of updating the node state, the information of the input nodes and edges is aggregated and updated, and the node label is output. The formula is expressed as:

[0016]

[0017]

[0018] In the formula, is the local transition function for updating the node state, is the local output function, is the state vector learned by the graph neural network iteration, is the node label;

[0019] In the continuous iteration of the graph neural network, the correlation between a node and its adjacent nodes is calculated to obtain strong correlation features in the time domain and strong correlation features in the frequency domain.

[0020] Furthermore, in the continuous iteration of the graph neural network, the association between a certain node and its adjacent nodes is calculated to obtain strong correlation features in the time domain and strong correlation features in the frequency domain, including:

[0021] set up and are the vectors constructed by superimposing the state vectors of all nodes, all output labels, the feature vectors of all nodes and edges, and the feature vectors of all nodes. The formula can be written in a more compact form as follows:

[0022]

[0023]

[0024] in, express No. Iterations, Represented by the eigenvector and the The state vector of the iteration is obtained by the global transfer function The state vector of all nodes in the iteration, and are the global transfer function and the global output function, respectively, and are the local transfer functions of all nodes. and local output function Stacking;

[0025] During the iteration process, nodes with similar states, nodes with complementary states, and the influence of each node on the overall graph neural network are determined. One of the nodes with similar states is selected, nodes with complementary states are merged into one node, and the node with the greatest influence on the overall graph neural network is used as a strong correlation feature.

[0026] Further, the extracting of the time domain strong correlation feature of the electromagnetic field signal, determining that a partial discharge signal is monitored based on the time domain strong correlation feature, and outputting a time domain waveform diagram includes:

[0027] Extracting the time-domain strong correlation feature from the electromagnetic field signal, and analyzing and matching the time-domain strong correlation feature to determine whether the electromagnetic field signal exceeds a first discharge threshold;

[0028] If yes, it is determined that a partial discharge signal is detected and a time domain waveform is output;

[0029] If not, the electromagnetic field signal is deleted.

[0030] Further, the extracting of the frequency domain strong correlation feature of the electromagnetic field signal, determining that a partial discharge signal is monitored based on the frequency domain strong correlation feature, and outputting a frequency domain waveform diagram includes:

[0031] Performing at least two fast Fourier transforms on the electromagnetic field signal to obtain frequency domain data of at least two frequency bands;

[0032] Extracting the frequency domain strong correlation features from the frequency domain data of each frequency band, and analyzing and matching the frequency domain strong correlation features to determine whether the frequency domain data of each frequency band exceeds the second discharge threshold;

[0033] If so, it is determined that a partial discharge signal is detected and the frequency domain waveform corresponding to each frequency band is normalized to obtain a frequency domain waveform diagram during discharge;

[0034] If not, the electromagnetic field signal is deleted.

[0035] Furthermore, the obtaining of a global waveform diagram of a partial discharge signal based on the time domain waveform diagram and the frequency domain waveform diagram comprises:

[0036] Converting the frequency domain waveform into a second time domain waveform during discharge;

[0037] The first time domain waveform diagram and the second time domain waveform diagram are combined to obtain a global domain waveform diagram of the partial discharge signal.

[0038] Furthermore, after obtaining the global waveform diagram of the partial discharge signal, the method further includes:

[0039] Based on the global waveform diagram of the partial discharge signal, the GIS discharge type and discharge point are determined.

[0040] In addition, the present invention also proposes a GIS partial discharge signal detection device, the device comprising:

[0041] Signal acquisition module, used to collect electromagnetic field signals inside GIS;

[0042] A strong correlation feature screening module is used to use the time-frequency domain features of the electromagnetic field signal as nodes of the graph neural network to perform correlation strength screening to obtain time-domain strong correlation features and frequency-domain strong correlation features, wherein the lines between the nodes represent the relationship between the features represented by the nodes;

[0043] A global waveform calculation module is used to extract the time domain strong correlation features of the electromagnetic field signal, determine that a local discharge signal is monitored based on the time domain strong correlation features, and output a first time domain waveform during discharge; extract the frequency domain strong correlation features of the electromagnetic field signal, determine that a local discharge signal is monitored based on the frequency domain strong correlation features, and output a frequency domain waveform during discharge; and obtain a global waveform of the local discharge signal based on the first time domain waveform and the frequency domain waveform.

[0044] Furthermore, the signal acquisition module includes a UHF signal acquisition antenna, an amplifying and filtering circuit and a multi-channel AD converter. The UHF signal acquisition antenna is installed on the opening of the GIS cylinder wall. The output of the UHF signal acquisition antenna is connected to the multi-channel AD converter via the amplifying and filtering circuit. The output of the multi-channel AD converter is connected to the global waveform calculation module.

[0045] Furthermore, the amplifying and filtering circuit includes a radio frequency gain circuit, a bandpass filter and a bandstop filter which are connected in sequence.

[0046] Furthermore, the global waveform calculation module adopts an FPGA chip, and the FPGA chip includes a storage unit, a frequency domain feature extraction unit, a time domain feature extraction unit and a waveform synthesis unit, wherein:

[0047] The storage unit is used to store the electromagnetic field signal;

[0048] The time domain feature extraction unit is used to extract the time domain strong correlation feature from the electromagnetic field signal, and output a time domain waveform graph after analyzing and matching the time domain strong correlation feature to determine that the electromagnetic field signal exceeds a first discharge threshold, otherwise, the electromagnetic field signal in the storage unit is deleted;

[0049] The frequency domain feature extraction unit is used to perform at least two fast Fourier transforms on the electromagnetic field signal to obtain frequency domain data of at least two frequency bands; extract the frequency domain strong correlation features from the frequency domain data of each frequency band, and output a frequency domain waveform diagram during discharge when it is determined that the frequency domain data of each frequency band exceeds a second discharge threshold after analyzing and matching the frequency domain strong correlation features; otherwise, the electromagnetic field signal in the storage unit is deleted.

[0050] Furthermore, the strongly associated feature screening module is specifically used for:

[0051] Complete feature map in graph neural network middle, is a node set, is the edge set, let Respectively represent nodes The eigenvectors of the edges, the eigenvectors of the nodes and the state vectors of its surrounding nodes and nodes The feature vectors of the surrounding nodes;

[0052] According to the local transfer function of updating the node state, the information of the input nodes and edges is aggregated and updated, and the node label is output. The formula is expressed as:

[0053]

[0054]

[0055] In the formula, is the local transition function for updating the node state, The local is the output function, is the state vector learned by the graph neural network iteration, is the node label;

[0056] In the continuous iteration of the graph neural network, the correlation between a node and its adjacent nodes is calculated to obtain strong correlation features in the time domain and strong correlation features in the frequency domain.

[0057] In addition, the present invention also proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the GIS partial discharge signal detection method as described above is implemented.

[0058] The advantages of the present invention are:

[0059] (1) The present invention substitutes the time-frequency domain features of multiple electromagnetic field signal detection into the graph neural network, and uses each time-frequency domain feature as a node of the graph neural network. The lines between the nodes represent the association between the features represented by the nodes. The graph neural network is iterated to perform association strength screening to obtain the features with the highest association, i.e., the strongly correlated features. Then, the local discharge signal is detected based on the screened strongly correlated features, which reduces the amount of calculation. However, since the correlation between the indicators is taken into account, the accuracy of the local discharge signal detection is not affected. Therefore, the present invention improves the accuracy of the local discharge signal detection while greatly reducing the amount of calculation.

[0060] (2) When the amount of collected signal data is large, if FFT calculation is performed directly, the processing speed is slow due to the large amount of data, which is not conducive to real-time signal detection and processing. Therefore, the present invention adopts an array processing method to divide the FFT results into at least two frequency bands for parallel processing, and calculate the correlation of the waveforms of at least two frequency bands respectively, so that the correlation determination basis can be increased to multiple, thereby greatly increasing the amount of information reflecting the relative deformation of the waveform, and at the same time, it can fully reflect the changes in the frequency response curve in different frequency intervals due to different degrees of discharge, and finally fuse the waveforms of multiple frequency bands. Therefore, the array processing mechanism of the present invention is a frequency domain segmentation and combination mechanism. Each part in the array only processes a part of the frequency domain, and then fuses them together to restore the complete frequency domain waveform. The mechanism disperses the noise signal superimposed on the frequency domain data to the segmented part, reducing the stacking of noise signals, which will help to achieve high-sensitivity detection of partial discharge UHF signals.

[0061] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of a GIS partial discharge signal detection method proposed in one embodiment of the present invention;

[0063] Figure 2 is a complete feature graph of a graph neural network in one embodiment of the present invention;

[0064] Figure 3 It is a schematic diagram of the neural network iteration process in one embodiment of the present invention;

[0065] Figure 4 is a schematic diagram of a calculation flow of a global waveform diagram of a partial discharge signal in one embodiment of the present invention;

[0066] Figure 5 It is a structural schematic diagram of a GIS partial discharge signal detection device proposed in one embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not 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.

[0068] like Figure 1As shown, the first embodiment of the present invention proposes a GIS partial discharge signal detection method, the method comprising the following steps:

[0069] S10, collecting electromagnetic field signals inside the GIS, using the time-frequency domain features of the electromagnetic field signals as nodes of the graph neural network to perform correlation strength screening, and obtaining time-domain strong correlation features and frequency-domain strong correlation features, wherein the lines between the nodes represent the relationship between the features represented by the nodes;

[0070] It should be noted that the electromagnetic field signals inside GIS can be collected using UHF sensors, UHF signal collection antennas, etc., and the collected electromagnetic field signals are time domain data.

[0071] S20, extracting the time domain strong correlation feature from the electromagnetic field signal, determining that a partial discharge signal is monitored based on the time domain strong correlation feature, and outputting a first time domain waveform diagram during discharge;

[0072] S30, extracting the frequency domain strong correlation feature from the electromagnetic field signal, and determining that a partial discharge signal is monitored based on the frequency domain strong correlation feature, and outputting a frequency domain waveform diagram during discharge;

[0073] S40. Obtain a global waveform diagram of a partial discharge signal based on the first time domain waveform diagram and the frequency domain waveform diagram.

[0074] It should be noted that there are many time-frequency domain indicators for detecting local discharge signals. If all time-frequency domain indicators are calculated to detect local discharge signals, the detection results are theoretically very accurate. However, if the detection and judgment are based on all time-frequency domain indicators, it will take up a large amount of computing resources. The computing resources of the current hardware cannot be calculated, and the accuracy cannot be met by randomly selecting certain features for calculation. This embodiment uses a neural network to screen out strongly correlated features for subsequent local discharge signal detection, thereby greatly reducing the amount of calculation and improving the accuracy of local discharge signal detection. As a further preferred technical solution, the time-frequency domain features include time domain features and frequency domain features, and the time domain features include margin, skewness, root mean square, shape factor, kurtosis, pulse factor and C index; the frequency domain features include mean frequency, frequency concentration, frequency center, root mean square frequency and main frequency band position change. Specifically, there are currently many time-frequency domain features for detecting local discharge signals, as shown in Table 1:

[0075] Table 1 Proposed time-frequency domain features

[0076]

[0077] Where: is the maximum absolute value (peak value), is the square root amplitude, is the skewness, Time domain waveform data The third-order center distance, Time domain waveform data The standard deviation of is the root mean square, For the Time domain data, is the total number of time domain data points, is the shape factor, For the deviation: is the kurtosis, is the pulse factor, is the time domain waveform data, is the mean frequency, is the amplitude of each frequency on the spectrum, is a fixed coefficient used to adjust the ratio in the formula. is the frequency concentration index 1, is the frequency concentration index 2, is the frequency concentration index 3, is the frequency center index, is the frequency of each point on the spectrum, is the frequency concentration index 4, is the root mean square frequency index, is the main frequency band position change index 1, is the main frequency band position change index 2, is the frequency concentration index 5, is the frequency concentration index 6, is the frequency concentration index 7, is the frequency concentration index 8, For the multiplication sign.

[0078] It should be noted that the indicators listed in Table 1 are common time-frequency domain indicators, some of which contain certain characteristics of UHF and are common UHF signal indicators. These characteristic indicators have more or less information that can be used for UHF signal detection, but this information may be repeated or complementary in multiple features. Therefore, this embodiment uses a graph neural network to filter out indicators with strong correlation among these characteristic indicators.

[0079] As a further preferred technical solution, in step S10, the time-frequency domain features of the electromagnetic field signal are used as nodes of a graph neural network to perform correlation strength screening to obtain time-domain strong correlation features and frequency-domain strong correlation features, including the following steps:

[0080] S11. Complete feature graph in graph neural network middle, is a node set, is the edge set, let Respectively represent nodes The eigenvectors of the edges, the eigenvectors of the nodes and the state vectors of its surrounding nodes and nodes The feature vectors of surrounding nodes;

[0081] It should be noted that the characteristic vector of a node is some inherent state of the node, which will not change during the iteration process of the graph neural network. Specifically in this embodiment, it is some parameter indicators in the time-frequency domain feature calculation formula of the electromagnetic field signal, such as standard deviation, peak value, etc. The characteristic vector of an edge refers to the connection between two nodes. Specifically in this embodiment, it refers to the same parameter indicators or similar processing processes in the calculation formula between nodes. Connecting these similar feature nodes helps the model learn the relationship and mutual influence between these features. The state vector refers to the continuously updated node vector obtained according to the operation during the iteration process; the node label is the target value used for supervised learning tasks. In this embodiment, each node represents a signal feature, and the corresponding label is used to represent the category and state of these signal features.

[0082] S12. According to the local transfer function of updating the node state, the information of the input nodes and edges is aggregated and updated, and the node label is output. The formula is expressed as:

[0083]

[0084]

[0085] In the formula, is the local transition function for updating the node state, is the output function, is the state vector learned by the graph neural network iteration, is the node label;

[0086] S13. In the continuous iteration of the graph neural network, the correlation between a node and its adjacent nodes is calculated to obtain strong correlation features in the time domain and strong correlation features in the frequency domain.

[0087] Specifically, the graph neural network GNN can calculate the connection between a node and its adjacent nodes. Through multiple iterations, it can also find the relationship between the adjacent nodes of the adjacent nodes and the current node. It is very suitable for screening strong correlation features. In this embodiment, all the features of the GIS partial discharge time-frequency domain signal (as shown in Table 1 above) are imported into the graph neural network as nodes in the network. The complete feature map constructed is as follows: Figure 2 As shown, Figure 2 Each node in represents a single feature, and the lines between nodes represent the relationship between the features represented by the nodes. Some features are highly correlated, such as and The Chinese Communist Party , while the relationship between some features is weak, so no connection is made. Blank nodes represent features that are not closely related to other features.

[0088] During the iteration process of the graph neural network, the feature vector of the node itself will not change with the network iteration; while the state vector between nodes will change with the iteration of the GNN network. Therefore, the graph neural network can be used to calculate the connection between a node and its adjacent nodes. Through multiple iterations, the connection between the adjacent nodes of the adjacent nodes and the current node can also be found. Features with strong correlation can be screened out from all the features in Table 1 for the detection of partial discharge signals.

[0089] As a further preferred technical solution, Figure 3 As shown, step S13: in the continuous iteration of the graph neural network, the correlation between a certain node and its adjacent nodes is calculated to obtain the time domain strong correlation feature and the frequency domain strong correlation feature, including:

[0090] set up and are vectors constructed by superimposing the state vectors of all nodes, all output labels, the feature vectors of all nodes and edges, and the feature vectors of all nodes, respectively, where:

[0091]

[0092]

[0093]

[0094]

[0095] The formula can be written in a more compact form as:

[0096]

[0097]

[0098] in, express No. Iterations, Represented by the eigenvector and the The state vector of the iteration is obtained by the global transfer function The state vector of all nodes in the iteration, and are the global transfer function and the global output function, respectively, and are the local transfer functions of all nodes. and local output function Stacking;

[0099] During the iteration process, nodes with similar states, nodes with complementary states, and the degree of influence of each node on the overall graph neural network are determined. Among the nodes with similar states, one of the nodes is selected, nodes with complementary states are merged into one node, and the node with the greatest influence on the overall graph neural network is taken as a strong correlation feature.

[0100] It should be noted that through the continuous iteration of graph neural networks, it can be found that the states of some nodes in the graph neural network are becoming more and more similar, the states of some nodes are complementary, and the influence of some nodes on the overall graph neural network becomes smaller. By selecting one of the similar nodes, combining the complementary nodes into one, and then selecting the node that has the greatest impact on the overall system, the strong correlation characteristics of the system are obtained.

[0101] It should be understood that the graph neural network used in this embodiment pre-constrains the model through a loss function, and finally uses a gradient descent algorithm to minimize the loss, thereby obtaining the optimal parameters of the network model and obtaining a trained graph neural network. This embodiment substitutes the features in Table 1 above into the graph neural network, performs correlation strength screening, and obtains the features with the highest correlation, that is, the results of the strong correlation features are the root mean square, pulse factor and margin index in the time domain, and the root mean square frequency index, main frequency band position change index 1 and main frequency band position change index 2 in the frequency domain.

[0102] As a further preferred technical solution, the step S20: extracting the time domain strong correlation feature of the electromagnetic field signal, determining that a partial discharge signal is monitored based on the time domain strong correlation feature, and outputting a time domain waveform diagram, comprises the following steps:

[0103] S21, extracting the time domain strong correlation feature of the electromagnetic field signal, and after analyzing and matching the time domain strong correlation feature, determining whether the electromagnetic field signal exceeds a first discharge threshold, if so, executing step S22, if not, executing step S23;

[0104] S22, determining that a partial discharge signal is monitored and outputting a time domain waveform diagram;

[0105] S23, deleting the electromagnetic field signal.

[0106] It should be noted that the first discharge threshold in this embodiment is an empirical value, so this embodiment does not specifically limit its value.

[0107] As a further preferred technical solution, the step S30: extracting the frequency domain strong correlation feature of the electromagnetic field signal, and determining that a partial discharge signal is monitored based on the frequency domain strong correlation feature, and outputting a frequency domain waveform diagram, comprises the following steps:

[0108] S31, performing at least two fast Fourier transforms on the electromagnetic field signal to obtain frequency domain data of at least two frequency bands;

[0109] S32, extracting the frequency domain strong correlation features from the frequency domain data of each frequency band, and after analyzing and matching the frequency domain strong correlation features, determining whether the frequency domain data of each frequency band exceeds the second discharge threshold, if so, executing step S33, if not, executing step S34;

[0110] S33, determining that a partial discharge signal is monitored and normalizing the frequency domain waveform corresponding to each frequency band to obtain a frequency domain waveform diagram during discharge;

[0111] S34, deleting the electromagnetic field signal.

[0112] Specifically, when the amount of collected signal data is large, if FFT calculation is performed directly, the processing speed is slow due to the large amount of data, which is not conducive to real-time signal detection and processing. Therefore, the present invention adopts array processing, such as Figure 4 As shown, the FFT results are divided into four frequency bands for parallel processing (the frequency bands are 200M-525M, 525M-850M, 850M-1175M, and 1175M-1500M, respectively). The correlations on the four frequency bands are calculated respectively, which can increase the basis for correlation determination to multiple, thereby greatly increasing the amount of information reflecting the relative deformation of the waveform, and at the same time fully reflecting the changes in the frequency response curve in different frequency intervals due to different degrees of discharge, and finally the waveforms of multiple frequency bands are fused. Therefore, the array processing mechanism of the present invention is a frequency domain segmentation and combination mechanism, in which each part of the array only processes a part of the frequency domain, and then they are fused together to restore the complete frequency domain waveform. The mechanism disperses the noise signal superimposed on the frequency domain data to the segmented part, reducing the stacking of noise signals, which will help to achieve high-sensitivity detection of local discharge UHF signals.

[0113] As a further preferred technical solution, the step S40: obtaining a global waveform diagram of the partial discharge signal based on the time domain waveform diagram and the frequency domain waveform diagram, comprises the following steps:

[0114] S41, converting the frequency domain waveform into a second time domain waveform during discharge;

[0115] It should be noted that, in this embodiment, the frequency domain waveform diagram is converted into a second time domain waveform diagram through IFFT.

[0116] S42: Combining the first time domain waveform diagram with the second time domain waveform diagram to obtain a global domain waveform diagram of a partial discharge signal.

[0117] As a further preferred technical solution, after obtaining the global waveform diagram of the partial discharge signal, the method further includes the following steps:

[0118] Based on the global waveform diagram of the partial discharge signal, the GIS discharge type and discharge point are determined.

[0119] It should be noted that the complete UHF time domain waveform collected by this embodiment can realize accurate calculation of the discharge position and discharge intensity.

[0120] In addition, if Figure 5 As shown, the second embodiment of the present invention provides a GIS partial discharge signal detection device, characterized in that the device comprises:

[0121] The signal acquisition module 10 is used to collect electromagnetic field signals inside the GIS;

[0122] A strong correlation feature screening module 20 is used to use the time-frequency domain features of the electromagnetic field signal as nodes of the graph neural network to perform correlation strength screening to obtain time-domain strong correlation features and frequency-domain strong correlation features, wherein the lines between the nodes represent the relationship between the features represented by the nodes;

[0123] The global waveform calculation module 30 is used to extract the time domain strong correlation features of the electromagnetic field signal, determine that a local discharge signal is monitored based on the time domain strong correlation features, and output a first time domain waveform during discharge; extract the frequency domain strong correlation features of the electromagnetic field signal, determine that a local discharge signal is monitored based on the frequency domain strong correlation features, and output a frequency domain waveform during discharge; and obtain a global waveform of the local discharge signal based on the first time domain waveform and the frequency domain waveform.

[0124] As a further preferred technical solution, the signal acquisition module includes a UHF signal acquisition antenna, an amplifying and filtering circuit and a multi-channel AD converter. The UHF signal acquisition antenna is installed on the opening of the GIS cylinder wall. The output of the UHF signal acquisition antenna is connected to the multi-channel AD converter via the amplifying and filtering circuit. The output of the multi-channel AD converter is connected to the global waveform calculation module.

[0125] It should be noted that in this embodiment, multiple signal acquisition modules can be set to achieve all-round acquisition of GIS electromagnetic field signals. For example, when two signal acquisition modules are set, the UHF signal acquisition antenna in each acquisition module is fixed to the connection terminal of the GIS tube wall opening, and the signal is transmitted to the amplification and filtering circuit through a coaxial cable. When measuring the UHF signal, the two signal sampling modules are placed at both ends of the monitored GIS line and fixed on the GIS tube wall through the reserved partial discharge detection hole.

[0126] As a further preferred technical solution, the amplifying and filtering circuit includes a radio frequency gain circuit, a bandpass filter and a bandstop filter connected in sequence.

[0127] It should be noted that the number of amplifying and filtering circuits provided in this embodiment corresponds to the number of signal acquisition modules. The output of each signal acquisition module is connected to an amplifying and filtering circuit. The amplifying and filtering circuit includes a RF gain module, a 300MHz~1500MHz band-pass filter, and an 800MHz~1000MHz band-stop filter. This embodiment performs band-pass and band-stop filtering according to the effective frequency band and common interference frequency band of the UHF signal in GIS, and is more suitable for UHF signal acquisition in GIS.

[0128] It should be noted that the design of this embodiment adopts an ultra-high-speed ADC (4G), which can meet the requirements of perfectly collecting discharge signals (up to 1.5 GHz).

[0129] As a further preferred technical solution, the global waveform calculation module adopts an FPGA chip, and the FPGA chip includes a storage unit, a frequency domain feature extraction unit, a time domain feature extraction unit and a waveform synthesis unit, wherein:

[0130] The storage unit is used to store the electromagnetic field signal;

[0131] The time domain feature extraction unit is used to extract the time domain strong correlation feature from the electromagnetic field signal, and output a time domain waveform graph after analyzing and matching the time domain strong correlation feature to determine that the electromagnetic field signal exceeds a first discharge threshold, otherwise, the electromagnetic field signal in the storage unit is deleted;

[0132] The frequency domain feature extraction unit is used to perform at least two fast Fourier transforms on the electromagnetic field signal to obtain frequency domain data of at least two frequency bands; extract the frequency domain strong correlation features from the frequency domain data of each frequency band, and output a frequency domain waveform diagram during discharge when it is determined that the frequency domain data of each frequency band exceeds a second discharge threshold after analyzing and matching the frequency domain strong correlation features; otherwise, the electromagnetic field signal in the storage unit is deleted.

[0133] It should be noted that the collected time domain data is completely sent to the high-performance storage module cache (Buffer). If it is determined that no partial discharge UHF signal is detected, the detected time domain data is deleted from the high-performance storage module (Buffer) to release storage space for the next collected time domain data.

[0134] As a further preferred technical solution, the strong correlation feature screening module is specifically used for:

[0135] Complete feature map in graph neural network middle, is a node set, is the edge set, let Respectively represent nodes The eigenvectors of the edges, the eigenvectors of the nodes and the state vectors of its surrounding nodes and nodes The feature vectors of the surrounding nodes;

[0136] According to the local transfer function of updating the node state, the information of the input nodes and edges is aggregated and updated, and the node label is output. The formula is expressed as:

[0137]

[0138]

[0139] In the formula, is the local transition function for updating the node state, is the output function, is the state vector learned by the graph neural network iteration, is the node label;

[0140] In the continuous iteration of the graph neural network, the correlation between a node and its adjacent nodes is calculated to obtain strong correlation features in the time domain and strong correlation features in the frequency domain.

[0141] It should be noted that other embodiments or implementation methods of the GIS partial discharge signal detection device of the present invention can refer to the above-mentioned method embodiments and will not be described in detail here.

[0142] In addition, the present invention also proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the GIS partial discharge signal detection method as described in the first embodiment above is implemented.

[0143] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0144] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0145] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0146] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0147] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A GIS partial discharge signal detection method, characterized in that: The method comprises: The electromagnetic field signal inside the GIS is collected, and the time-frequency domain characteristics of the electromagnetic field signal are used as nodes of the graph neural network for correlation strength screening to obtain strong correlation features in the time domain and strong correlation features in the frequency domain. The lines between the nodes represent the relationship between the features represented by the nodes, including the complete feature map of the graph neural network. middle, is a node set, is the edge set, let Respectively represent nodes The eigenvectors of the edges, the eigenvectors of the nodes and the state vectors of its surrounding nodes and nodes Feature vectors of surrounding nodes; aggregate and update the input node and edge information according to the local transfer function that updates the node state, and output the node label: , where is the local transition function for updating the node state, is the output function, is the state vector learned by the graph neural network iteration, , is the node label; in the continuous iteration of the graph neural network, the correlation between a node and its adjacent nodes is calculated to obtain the strong correlation features in the time domain and the strong correlation features in the frequency domain; Extracting the time-domain strong correlation feature from the electromagnetic field signal, determining that a partial discharge signal is monitored based on the time-domain strong correlation feature, and outputting a first time-domain waveform diagram during discharge; Extracting the frequency domain strong correlation feature from the electromagnetic field signal, determining that a partial discharge signal is monitored based on the frequency domain strong correlation feature, and outputting a frequency domain waveform diagram during discharge; Based on the first time domain waveform diagram and the frequency domain waveform diagram, a full domain waveform diagram of the partial discharge signal is obtained.

2. The GIS partial discharge signal detection method according to claim 1, characterized in that: The time-frequency domain features include time domain features and frequency domain features. The time domain features include margin, skewness, root mean square, mean, variance, peak-to-peak value, shape factor, kurtosis, impulse factor and C index; the frequency domain features include mean frequency, frequency concentration, frequency center, root mean square frequency and main frequency band position change.

3. The GIS partial discharge signal detection method according to claim 1, characterized in that: In the continuous iteration of the graph neural network, the association between a node and its adjacent nodes is calculated to obtain strong correlation features in the time domain and strong correlation features in the frequency domain, including: set up and are the vectors constructed by superimposing the state vectors of all nodes, all output labels, the feature vectors of all nodes and edges, and the feature vectors of all nodes. The formula can be written in a more compact form as follows: in, express No. Iterations, Represented by the eigenvector and the The state vector of the iteration is obtained by the global transfer function The state vector of all nodes in the iteration, and are the global transfer function and the global output function, respectively, and are the local transfer functions of all nodes. and local output function Stacking; During the iteration process, nodes with similar states, nodes with complementary states, and the influence of each node on the overall graph neural network are determined. One of the nodes with similar states is selected, nodes with complementary states are merged into one node, and the node with the greatest influence on the overall graph neural network is used as a strong correlation feature.

4. The GIS partial discharge signal detection method according to claim 1, characterized in that: The extracting of the time domain strong correlation feature of the electromagnetic field signal, determining that a partial discharge signal is monitored based on the time domain strong correlation feature, and outputting a time domain waveform diagram includes: Extracting the time-domain strong correlation feature from the electromagnetic field signal, and analyzing and matching the time-domain strong correlation feature to determine whether the electromagnetic field signal exceeds a first discharge threshold; If yes, it is determined that a partial discharge signal is detected and a time domain waveform is output; If not, the electromagnetic field signal is deleted.

5. The GIS partial discharge signal detection method according to claim 1, characterized in that: The extracting of the frequency domain strong correlation feature of the electromagnetic field signal, determining that a partial discharge signal is monitored based on the frequency domain strong correlation feature, and outputting a frequency domain waveform diagram includes: Performing at least two fast Fourier transforms on the electromagnetic field signal to obtain frequency domain data of at least two frequency bands; Extracting the frequency domain strong correlation features from the frequency domain data of each frequency band, and analyzing and matching the frequency domain strong correlation features to determine whether the frequency domain data of each frequency band exceeds the second discharge threshold; If so, it is determined that a partial discharge signal is detected and the frequency domain waveform corresponding to each frequency band is normalized to obtain a frequency domain waveform diagram during discharge; If not, the electromagnetic field signal is deleted.

6. The GIS partial discharge signal detection method according to claim 1, characterized in that: The method of obtaining a global waveform diagram of a partial discharge signal based on the time domain waveform diagram and the frequency domain waveform diagram comprises: Converting the frequency domain waveform into a second time domain waveform during discharge; The first time domain waveform diagram and the second time domain waveform diagram are combined to obtain a global domain waveform diagram of the partial discharge signal.

7. The GIS partial discharge signal detection method according to claim 1, characterized in that: After obtaining the global waveform diagram of the partial discharge signal, the method further includes: Based on the global waveform diagram of the partial discharge signal, the GIS discharge type and discharge point are determined.

8. A GIS partial discharge signal detection device, characterized in that: The device comprises: Signal acquisition module, used to collect electromagnetic field signals inside GIS; A strong correlation feature screening module is used to use the time-frequency domain features of the electromagnetic field signal as nodes of the graph neural network to screen the correlation strength, and obtain time-domain strong correlation features and frequency-domain strong correlation features, wherein the lines between the nodes represent the relationship between the features represented by the nodes, including the complete feature graph of the graph neural network. middle, is a node set, is the edge set, let Respectively represent nodes The eigenvectors of the edges, the eigenvectors of the nodes and the state vectors of its surrounding nodes and nodes Feature vectors of surrounding nodes; aggregate and update the input node and edge information according to the local transfer function that updates the node state, and output the node label: , where is the local transition function for updating the node state, is the output function, is the state vector learned by the graph neural network iteration, , is the node label; in the continuous iteration of the graph neural network, the correlation between a node and its adjacent nodes is calculated to obtain the strong correlation features in the time domain and the strong correlation features in the frequency domain; A global waveform calculation module is used to extract the time domain strong correlation features of the electromagnetic field signal, determine that a local discharge signal is monitored based on the time domain strong correlation features, and output a first time domain waveform during discharge; extract the frequency domain strong correlation features of the electromagnetic field signal, determine that a local discharge signal is monitored based on the frequency domain strong correlation features, and output a frequency domain waveform during discharge; and obtain a global waveform of the local discharge signal based on the first time domain waveform and the frequency domain waveform.

9. The GIS partial discharge signal detection device according to claim 8, characterized in that: The signal acquisition module includes a UHF signal acquisition antenna, an amplifying and filtering circuit and a multi-channel AD converter. The UHF signal acquisition antenna is installed on the opening of the GIS cylinder wall. The output of the UHF signal acquisition antenna is connected to the multi-channel AD converter through the amplifying and filtering circuit. The output of the multi-channel AD converter is connected to the global waveform calculation module.

10. The GIS partial discharge signal detection device according to claim 9, characterized in that: The amplifying and filtering circuit comprises a radio frequency gain circuit, a bandpass filter and a bandstop filter which are connected in sequence.

11. The GIS partial discharge signal detection device according to claim 8, characterized in that: The global waveform calculation module adopts an FPGA chip, which includes a storage unit, a frequency domain feature extraction unit, a time domain feature extraction unit and a waveform synthesis unit, wherein: The storage unit is used to store the electromagnetic field signal; The time domain feature extraction unit is used to extract the time domain strong correlation feature from the electromagnetic field signal, and output a time domain waveform graph after analyzing and matching the time domain strong correlation feature to determine that the electromagnetic field signal exceeds a first discharge threshold, otherwise, the electromagnetic field signal in the storage unit is deleted; The frequency domain feature extraction unit is used to perform at least two fast Fourier transforms on the electromagnetic field signal to obtain frequency domain data of at least two frequency bands; extract the frequency domain strong correlation features from the frequency domain data of each frequency band, and output a frequency domain waveform diagram during discharge when it is determined that the frequency domain data of each frequency band exceeds a second discharge threshold after analyzing and matching the frequency domain strong correlation features; otherwise, the electromagnetic field signal in the storage unit is deleted.

12. 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 7 is implemented.

Images