Relay protection reliability evaluation method, system and equipment based on intelligent oscillograph and storage medium

Abstract

The invention discloses a relay protection reliability evaluation method, system and equipment based on an intelligent oscillograph and a storage medium, and relates to the technical field of power system relay protection, and the method comprises the steps: constructing a multi-source data collection network at each node of a power distribution network, collecting multi-source recording data, carrying out the time synchronization marking, and constructing a multi-source time sequence association data set; extracting a fault feature vector based on the multi-source time sequence association data set; based on the fault feature vectors, constructing a power distribution network topology model through a graph neural network, establishing a fault propagation relationship between nodes, and constructing a multi-source recording data collaborative analysis model by adopting a recurrent neural network; inputting the real-time recording data into the collaborative analysis model, calculating an action time sequence correlation index and a fault propagation influence index, evaluating the selective cooperation degree of the protection device, and generating an evaluation report; according to the invention, the accuracy and real-time performance of reliability evaluation of the relay protection device are improved, and the cooperative cooperation capability between the protection devices is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of relay protection of power systems, and in particular to a relay protection reliability evaluation method, system, device and storage medium based on an intelligent oscilloscope. Background Art

[0002] Relay protection is crucial for the safe and stable operation of power systems. Intelligent oscilloscope recorders, as key data acquisition devices in distribution networks, can record various electrical quantities and protective device operation information during power system faults. As distribution networks continue to expand and become more intelligent, relay protection reliability assessment technology based on intelligent oscilloscope recorders is gaining widespread application. By analyzing recorded data to evaluate the performance of protective device operation, they can optimize and adjust the selective coordination of protective devices.

[0003] The existing technology has the following deficiencies: First, the data collected by traditional recorders lack a unified time base, making it difficult to achieve accurate timing correlation of the recorded data of multiple nodes, affecting the accuracy of the timing analysis of the protection device action; second, the existing recorded data analysis method is mainly based on the fault characteristics of a single node, and does not fully consider the distribution network topology and fault propagation characteristics, and cannot effectively evaluate the coordination effect between protection devices; third, the reliability assessment of protection devices lacks systematicity and real-timeness, making it difficult to timely discover and solve the selective coordination problem of protection devices, affecting the timeliness and accuracy of distribution network fault handling. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: the lack of time series correlation in the data collection of the intelligent recorder, the single analysis model of the recorded data, and the insufficient systematic and real-time technical problems in the reliability evaluation of the protection device.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for evaluating relay protection reliability based on an intelligent oscilloscope, comprising:

[0008] Build a multi-source data acquisition network at each node of the distribution network to collect multi-source recorded data and perform time synchronization marking to build a multi-source time series correlation data set;

[0009] Extract fault feature vectors based on multi-source time series correlation datasets;

[0010] Based on the fault feature vector, a distribution network topology model is constructed through a graph neural network, and the fault propagation relationship between nodes is established. A multi-source recording data collaborative analysis model is constructed using a recursive neural network.

[0011] The real-time recorded data is input into the collaborative analysis model to calculate the action timing correlation index and fault propagation impact index, evaluate the degree of selective coordination of the protection device and generate an evaluation report.

[0012] As a preferred solution for the relay protection reliability evaluation method based on intelligent oscilloscope, the following methods are proposed:

[0013] The method of constructing a multi-source data acquisition network at each node of the distribution network, collecting multi-source recorded wave data and performing time synchronization marking, and constructing a multi-source time series correlation data set includes:

[0014] Intelligent recorders are installed at each node of the distribution network to collect multi-source recording data including primary side voltage signals, primary side current signals, secondary side protection device action signals, and secondary side protection device status signals.

[0015] As a preferred solution for the relay protection reliability evaluation method based on intelligent oscilloscope, the following methods are proposed:

[0016] The multi-source data acquisition network is constructed at each node of the distribution network to collect multi-source recorded wave data and perform time synchronization marking, and the multi-source time series correlation data set is constructed, which further includes:

[0017] Multi-source recorded data are time-synchronized and marked, a unified microsecond-level time reference is established, and a multi-source time series correlation data set is formed.

[0018] The beneficial effect of this preferred technical solution is that it synchronizes and marks multi-source recorded data and establishes a unified microsecond-level time base, ensuring the consistency and accuracy of different types of data in the temporal dimension. This makes subsequent analysis of correlations between multi-source data more reliable. In particular, when evaluating the correlation of relay protection device operation timing, precise time synchronization can avoid misjudgments caused by time errors, thereby improving the accuracy and effectiveness of relay protection reliability assessments.

[0019] As a preferred solution for relay protection reliability evaluation based on intelligent oscilloscope, the following methods are proposed:

[0020] Extracting the fault feature vector based on the multi-source time series correlation data set includes:

[0021] Eliminate noise interference from multi-source time series correlation data sets through wavelet transform;

[0022] The fundamental component of a fixed number of sampling points within the period is extracted from the denoised voltage signal. The fundamental amplitude, phase angle and distortion rate are obtained by sliding calculation of adjacent data windows to form a voltage feature vector.

[0023] The current mutation amount, mutation rate and effective value of the de-noised current signal are calculated to obtain time domain features, and the wavelet energy entropy and singular value decomposition are extracted to obtain frequency domain features. The time domain features and frequency domain features are combined to form a current feature vector.

[0024] Based on the noise-reduced protection signal, an action sequence feature vector including the action type, action time and duration of the protection device is constructed. A state transition matrix is ​​established, and the state transition probability and state duration are extracted to form the protection signal feature vector.

[0025] The beneficial effects of this preferred technical solution are: by performing noise reduction processing on multi-source time-series correlation data sets through wavelet transform, noise interference can be effectively removed and data quality can be improved. Appropriate feature extraction methods are used for different types of signals to accurately capture the key features in voltage, current and protection signals. For example, the fundamental amplitude, phase angle and distortion rate in the voltage feature vector can reflect the stability of the voltage; the current feature vector combines the time domain and frequency domain features to more comprehensively reflect the changing characteristics of the current; the protection signal feature vector records the action information and state transition of the protection device in detail. The extraction of these feature vectors provides effective input for the subsequent construction of accurate distribution network topology models and collaborative analysis models, which helps to improve the accuracy and reliability of relay protection reliability assessment.

[0026] As a preferred solution for relay protection reliability evaluation based on intelligent oscilloscope, the following methods are proposed:

[0027] The method of constructing a distribution network topology model based on fault feature vectors and establishing fault propagation relationships between nodes through a graph neural network and constructing a multi-source waveform data collaborative analysis model using a recursive neural network includes:

[0028] A distribution network topology model is constructed based on a graph neural network, fault feature vectors are mapped to the distribution network topology model, and the fault propagation relationship between nodes is established;

[0029] Recursive neural networks are used to process the fault propagation relationship between nodes and a collaborative analysis model for multi-source recorded data is constructed.

[0030] The beneficial effects of this preferred technical solution are as follows: the graph neural network can effectively represent the topological structure of the distribution network, map the fault feature vector to the topological structure model, and can intuitively display the fault propagation relationship between nodes, providing a clear model for analyzing the diffusion path and impact range of the fault. Recursive neural networks are good at processing sequence data and time dependencies. By using them to process the fault propagation relationship between nodes, it can capture the dynamic propagation process of faults in the distribution network. By constructing a collaborative analysis model for multi-source recording data, it is possible to comprehensively consider multi-source data and fault propagation relationships, improve the comprehensiveness and accuracy of relay protection reliability assessment, and more effectively predict the occurrence and development of faults.

[0031] As a preferred solution for relay protection reliability evaluation based on intelligent oscilloscope, the following methods are proposed:

[0032] The inputting of real-time recorded wave data into the collaborative analysis model, calculating the action timing correlation index and the fault propagation impact index, evaluating the selectivity of the protection device and generating an evaluation report includes:

[0033] The multi-source recorded data collected in real time is input into the multi-source recorded data collaborative analysis model to calculate the action timing correlation index and fault propagation impact index of the relay protection device.

[0034] As a preferred solution for relay protection reliability evaluation based on intelligent oscilloscope, the following methods are proposed:

[0035] The inputting of real-time recorded wave data into the collaborative analysis model, calculating the action timing correlation index and the fault propagation impact index, evaluating the selectivity coordination degree of the protection device and generating an evaluation report also includes:

[0036] Evaluate the selective coordination degree of relay protection devices based on action sequence correlation indicators and fault propagation impact indicators;

[0037] Generate a coordinated evaluation report for relay protection devices, including evaluation results of coordination timing, fault isolation range, and protection blocking relationships between protection devices.

[0038] In a second aspect, an embodiment of the present invention provides a relay protection reliability assessment system based on an intelligent oscilloscope, comprising:

[0039] A multi-source time series correlation data set construction module is used to build a multi-source data acquisition network at each node of the distribution network, collect multi-source recorded wave data, perform time synchronization marking, and construct a multi-source time series correlation data set;

[0040] Fault feature vector extraction module, used to extract fault feature vectors based on multi-source time series correlation data sets;

[0041] The analysis and modeling module is used to construct a distribution network topology model based on fault feature vectors through a graph neural network and establish the fault propagation relationship between nodes. It also uses a recursive neural network to build a collaborative analysis model for multi-source recorded data.

[0042] The evaluation module is used to input real-time recorded data into the collaborative analysis model, calculate the action timing correlation index and fault propagation impact index, evaluate the degree of selective coordination of the protection device and generate an evaluation report.

[0043] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0044] memory and processor;

[0045] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the one or more programs are executed by the one or more processors, the one or more processors implement the relay protection reliability assessment method based on the intelligent recorder as described in any embodiment of the present invention.

[0046] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the relay protection reliability assessment method based on an intelligent oscilloscope.

[0047] Beneficial effects of the present invention: The present invention can more accurately analyze fault characteristics, propagation relationships and action timings of protection devices through multi-source recording data acquisition and collaborative analysis models, thereby more accurately evaluating the reliability of relay protection; utilize intelligent recorders and high-speed fiber-optic communication networks to achieve rapid acquisition and transmission of multi-source recording data, and through advanced algorithms such as graph neural networks and recursive neural networks, achieve automated collaborative analysis, thereby improving evaluation efficiency; not only can the action timing correlation of protection devices be evaluated, but also the impact of fault propagation can be evaluated, and a comprehensive evaluation report containing the coordination timing between protection devices, fault isolation range, and protection locking relationship can be generated, thereby improving the comprehensiveness of the evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 any creative labor.

[0049] Figure 1 It is an overall flow chart of the relay protection reliability evaluation method based on the intelligent recorder described in the present invention. DETAILED DESCRIPTION

[0050] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0051] Example 1, with reference to Figure 1 , which is the first embodiment of the present invention, provides a method for evaluating the reliability of relay protection based on an intelligent oscilloscope, comprising:

[0052] S1: Build a multi-source data acquisition network at each node of the distribution network, collect multi-source recorded data, perform time synchronization marking, and construct a multi-source time series correlation data set;

[0053] S2: Extract fault feature vectors based on multi-source time series correlation datasets;

[0054] S3: Based on the fault feature vector, a distribution network topology model is constructed through a graph neural network, and the fault propagation relationship between nodes is established. A multi-source recording data collaborative analysis model is constructed using a recursive neural network.

[0055] S4: Input the real-time recorded data into the collaborative analysis model, calculate the action timing correlation index and fault propagation impact index, evaluate the selectivity of the protection device and generate an evaluation report.

[0056] It should be noted that, through steps S1-S4, this embodiment implements a series of steps such as building a multi-source data acquisition network at each node of the distribution network, extracting fault feature vectors, building a distribution network topology model and a collaborative analysis model of multi-source recording data, and calculating relevant indicators and generating an evaluation report, thereby achieving a comprehensive and accurate evaluation of the reliability of relay protection. This method makes full use of the information of multi-source recording data, combines advanced graph neural network and recursive neural network technology, and can accurately capture the fault characteristics and fault propagation relationships in the distribution network, and effectively evaluate the selective coordination degree of the protection device. In practical applications, this method helps to timely discover potential problems in the distribution network relay protection system, provide strong guarantees for the safe and stable operation of the power system, reduce the losses and impacts caused by faults, has significant economic value and social benefits, and has good prospects for promotion and application.

[0057] Example 2, reference Figure 1 , which is an embodiment of the present invention, provides a relay protection reliability evaluation method based on an intelligent oscilloscope based on the previous embodiment, including:

[0058] In this embodiment, in the above step S1, a multi-source data acquisition network is constructed at each node of the distribution network, multi-source recorded wave data is collected and time-synchronized, and a multi-source time series correlation data set is constructed, which includes:

[0059] Intelligent recorders are installed at each node of the distribution network to realize data transmission through a high-speed fiber-optic communication network. Multi-source recording data including primary-side voltage signal, primary-side current signal, secondary-side protection device action signal and secondary-side protection device status signal are collected. Multi-source recording data are time-synchronized and marked to establish a unified microsecond-level time base, thus forming a multi-source time-series correlation data set.

[0060] A multi-source wave recording data acquisition network is constructed, and an intelligent wave recorder is installed at each node of the distribution network. The intelligent wave recorder includes an analog quantity acquisition channel and a digital quantity acquisition channel. The intelligent wave recorder realizes data transmission through a high-speed fiber optic communication network. The high-speed fiber optic communication network has a dual-ring network structure. The backbone ring network of the high-speed fiber optic communication network adopts 10 Gigabit Ethernet technology and supports IEEE1588v2 precision time protocol. The intelligent wave recorder is connected to the backbone ring network through a fiber optic transceiver and uses VLAN technology to divide the service domain.

[0061] Specifically, the primary side voltage signal and the primary side current signal are collected through the analog acquisition channel, wherein the primary side voltage signal is collected by an electronic voltage transformer with a bandwidth of 10kHz, and the primary side current signal is collected by a Rogowski coil with a measurement range of 5mA to 2000A; the secondary side protection device action signal and the secondary side protection device status signal are collected through the digital acquisition channel, wherein the secondary side protection device action signal is collected by an optocoupler isolator, and the secondary side protection device status signal is collected by the RS-485 interface.

[0062] Furthermore, a unified microsecond-level time base is established. This unified microsecond-level time base is achieved through a three-level time synchronization system, which includes a GPS-based timing system, a communication network clock based on the IEEE1588v2 protocol, and an internal clock synchronization circuit of an intelligent wave recorder using phase-locked loop technology. Multi-source wave recording data is time-synchronized and marked based on the unified microsecond-level time base. The multi-source wave recording data that has been time-synchronized and marked is partitioned and stored in a distributed time series database according to the time and space dimensions.

[0063] Perform data quality assessment on the stored multi-source recording data, and determine the reliability of the multi-source recording data through data integrity verification, mutation detection and trend analysis; align the multi-source recording data that have passed the reliability assessment based on timestamps to form a multi-source time series correlation data set.

[0064] In another possible implementation, existing surveillance cameras in the distribution network can be used to build a video capture network. High-definition surveillance cameras are installed at key nodes in the distribution network, such as substations and towers, to capture real-time images of equipment appearance and operational status videos. Using image recognition and video analysis technologies, information such as equipment appearance anomalies (such as surface damage or smoke) and indicator light status can be extracted from the video data and included as part of the multi-source recording data. Furthermore, to ensure time synchronization of the video data, the cameras can be calibrated using the Network Time Protocol (NTP).

[0065] In this embodiment, extracting the fault feature vector based on the multi-source time series correlation data set in the above step S2 includes:

[0066] Perform data preprocessing on multi-source time series correlation data sets, eliminate noise interference through wavelet transform, and extract fault feature vectors;

[0067] Specifically, the primary side voltage signal in the multi-source time series correlation data set is mapped to the zero to one interval using the maximum-minimum value normalization method to obtain normalized voltage data. The primary side current signal in the multi-source time series correlation data set is normalized by using the Z-score normalization method to eliminate the dimension effect to obtain normalized current data. The secondary side protection device action signal and secondary side protection device status signal in the multi-source time series correlation data set are digitized using the one-hot encoding method to obtain normalized protection data.

[0068] The normalized voltage data, normalized current data, and normalized protection data are de-noised using wavelet transform. The db4 wavelet basis function with time-frequency localization is used for five-layer wavelet decomposition to obtain high-frequency detail coefficients and low-frequency approximation coefficients. The high-frequency detail coefficients are subjected to adaptive threshold processing using a soft thresholding method based on the Stein unbiased risk estimation criterion. The processed high-frequency detail coefficients and low-frequency approximation coefficients are then reconstructed using wavelet transform to obtain the denoised voltage, current, and protection signals.

[0069] The discrete Fourier transform algorithm is used to extract the fundamental components of the 32 sampling points in the cycle of the denoised voltage signal, and the fundamental amplitude, phase angle and distortion rate are obtained by sliding calculation of adjacent data windows to form the voltage feature vector; the current mutation amount, mutation rate and effective value of the denoised current signal are calculated to obtain the time domain characteristics, and the wavelet energy entropy and singular value decomposition are extracted to obtain the frequency domain characteristics, and the time domain characteristics and frequency domain characteristics are combined to form the current feature vector; based on the denoised protection signal, an action sequence feature vector including the action type, action time and duration of the protection device is constructed; based on the denoised protection signal, a state transition matrix is ​​established, and the state transition probability and state duration are extracted to form the protection signal feature vector.

[0070] In another possible implementation, principal component analysis (PCA) can be used to extract features from the primary-side voltage and current signals in a multi-source time-series correlation dataset. First, the data from multiple sampling points of the voltage and current signals is combined into a feature matrix. The PCA algorithm then performs dimensionality reduction on this matrix to extract the principal components that represent the majority of the original data. These principal components are then used as part of the fault feature vector. This method effectively reduces the data dimension and computational complexity.

[0071] In this embodiment, in step S3, based on the fault feature vector, a distribution network topology model is constructed using a graph neural network and a fault propagation relationship between nodes is established. A multi-source recording data collaborative analysis model is constructed using a recursive neural network, including:

[0072] A distribution network topology model is constructed based on a graph neural network, fault feature vectors are mapped to the distribution network topology model, and the fault propagation relationship between nodes is established;

[0073] Use recursive neural networks to process fault propagation relationships between nodes and build a collaborative analysis model for multi-source waveform data;

[0074] Specifically, a distribution network topology model is constructed based on a graph neural network, and a distribution network topology graph containing a node set and an edge set is established. The node set includes bus nodes, transformer nodes, and circuit breaker nodes in the distribution network, and the edge set represents the physical connection relationship between adjacent nodes in the node set. A node static feature vector containing node type information and rated parameter information is assigned to each node in the node set, and an adjacency matrix describing the connection relationship between nodes is constructed based on the edge set. A graph attention network is used to extract features from the distribution network topology graph, and the attention coefficient between adjacent nodes in the node set is calculated through a multi-head attention mechanism. The node static feature vector is weighted and aggregated based on the attention coefficient to obtain a node dynamic feature vector containing local structural information and global information.

[0075] The fault feature vector is mapped to the distribution network topology model, and the fault feature vector is concatenated with the node dynamic feature vector to obtain an enhanced node feature vector. A multi-layer perceptron is used to process the enhanced node feature vectors and corresponding edge physical parameters of adjacent nodes, calculate the fault propagation intensity between nodes, and construct a fault propagation matrix.

[0076] Establish fault propagation relationships between nodes, construct a gated recurrent unit network based on the fault propagation matrix, and use the fault feature vector and fault propagation intensity as input features of the gated recurrent unit network. Update gates are used to control the degree of retention of historical information, and reset gates are used to adjust the impact of historical states on current calculations, thereby obtaining a hidden state vector that reflects the dynamic propagation characteristics of the fault.

[0077] A recursive neural network is used to process the fault propagation relationship between nodes. A long short-term memory network is used to perform time series modeling on the hidden state vector. The input gate, forget gate, and output gate are used to control the flow of information, capture the long-term dependency of fault propagation, and output the fault propagation time series feature vector.

[0078] A collaborative analysis model for multi-source recorded data is constructed. The fault propagation time series feature vector is input into a decoder network consisting of a fully connected layer and a softmax layer. The fault propagation time series feature vector is mapped to the feature space through the fully connected layer, and the probability distribution of the fault type, fault location, and fault impact range is output through the softmax layer. The specific location of the fault, the type of fault, and the impact range of the fault are determined based on the probability distribution.

[0079] In another possible implementation, a hypergraph neural network can be used when constructing a distribution network topology model. Hypergraphs can more flexibly represent complex relationships between nodes. They can represent not only binary connection relationships between nodes, but also multivariate relationships between multiple nodes. In a hypergraph, the node set and the hyperedge set represent the various association relationships between device nodes and devices in the distribution network (such as electrical connections, communication connections, etc.). By extracting features from the hypergraph through a hypergraph neural network, the topological structure information of the distribution network can be captured more comprehensively. The fault feature vector is mapped into the hypergraph model to establish a more complex fault propagation relationship between nodes.

[0080] In another possible implementation, a combined model of a gated recurrent unit (GRU) and a convolutional neural network (CNN) can be used to construct a collaborative analysis model for multi-source waveform data. First, a CNN is used to extract spatial features from the fault feature vector, mining the local feature information within the fault signature. The extracted features are then fed into a GRU for temporal modeling. The GRU effectively captures the dynamic characteristics and long-term dependencies of fault propagation through update and reset gates. Finally, the GRU output is fed into a fully connected layer and a softmax layer to output a probability distribution of the fault type, location, and impact range.

[0081] In this embodiment, in step S4, the real-time recorded data is input into the collaborative analysis model, the action sequence correlation index and the fault propagation impact index are calculated, the selective coordination degree of the protection device is evaluated, and the evaluation report is generated, which includes:

[0082] Input the multi-source recorded data collected in real time into the multi-source recorded data collaborative analysis model to calculate the action sequence correlation index and fault propagation impact index of the relay protection device;

[0083] Evaluate the selective coordination degree of relay protection devices based on action sequence correlation indicators and fault propagation impact indicators;

[0084] Generate a coordinated evaluation report for relay protection devices, including evaluation results of coordination timing, fault isolation range, and protection blocking relationships between protection devices.

[0085] Specifically, multi-source waveform data from the distribution network is collected in real time. The multi-source waveform data includes fault current waveforms, fault voltage waveforms, protection device action information, and circuit breaker status information. The multi-source waveform data is input into a pre-trained multi-source waveform data collaborative analysis model, which is built based on graph neural networks and recursive neural networks.

[0086] Extract the action sequence information of the protection device based on the output results of the multi-source recording data collaborative analysis model. The action sequence information includes the action time, action type and duration. Calculate the time difference between the protection devices based on the action time. Calculate the time consistency component based on the time difference and the preset time attenuation coefficient. Calculate the action type similarity component based on the action type. Multiply the time consistency component by the action type similarity component to obtain the action timing correlation.

[0087] The dynamic time warping algorithm is used to process the action sequences of different protection devices. The dynamic time warping algorithm obtains the action sequence similarity by recursively calculating the minimum cumulative distance between sequence elements. The action sequence correlation index is obtained by weighted combination of the action sequence association degree and the action sequence similarity.

[0088] Based on the collaborative analysis model of multi-source recorded data, the fault propagation path is identified, and the ratio of the number of nodes affected by the fault to the total number of nodes in the distribution network is calculated to obtain the fault isolation efficiency. The fault propagation suppression rate is calculated based on the fault propagation intensity before and after the protection device is activated. The regional impact is calculated based on the node importance weight, fault diffusion distance, and affected duration. The fault isolation efficiency, fault propagation suppression rate, and regional impact are weighted and combined to obtain the fault propagation impact index.

[0089] Receive action timing correlation indicators and fault propagation impact indicators. The action timing correlation indicators include action timing correlation and action sequence similarity. The fault propagation impact indicators include fault isolation efficiency, fault propagation suppression rate, and regional impact.

[0090] The action sequence correlation degree and the deviation value of the actual action time interval of the protection device are weightedly combined to obtain the action sequence evaluation coefficient. The time selectivity evaluation value is calculated based on the action sequence evaluation coefficient and the action sequence similarity. The directional selectivity evaluation value is calculated based on the weighted combination of the fault isolation efficiency, the fault propagation suppression rate and the regional influence.

[0091] Calculate the current setting difference of adjacent protection devices, and use the ratio of the current setting difference to the difference between the maximum and minimum short-circuit currents at the fault point as the current selectivity evaluation value; construct a selectivity coordination matrix based on the time selectivity evaluation value, the direction selectivity evaluation value, and the current selectivity evaluation value, and calculate the selectivity coordination degree of the protection device based on the selectivity coordination matrix;

[0092] Obtain the blocking relationship information between protection devices and construct a blocking logic evaluation matrix; analyze the integrity of the blocking relationship based on the blocking logic evaluation matrix to obtain the blocking integrity index; calculate the blocking timing rationality index based on the time difference between the blocking signal taking effect and the protection action time; obtain the blocking reliability index by statistically analyzing the blocking success rate in historical operation data; and obtain the blocking scheme evaluation result by weighted combination of the blocking integrity index, blocking timing rationality index, and blocking reliability index.

[0093] In another possible implementation, the concept of information entropy can be introduced when calculating the action sequence relevance index. The action sequence information of the protective device is encoded, and the information entropy of the action sequence is calculated. Information entropy can reflect the uncertainty and complexity of the action sequence. Information entropy is used as a new characteristic component and weightedly combined with the action sequence relevance and action sequence similarity to obtain a more comprehensive action sequence relevance index. This method can more accurately assess the timing rationality of the protective device's actions.

[0094] Example 3. The above is a schematic scheme of the relay protection reliability assessment method based on an intelligent oscilloscope in this embodiment. It should be noted that the technical scheme of the relay protection reliability assessment system based on an intelligent oscilloscope and the technical scheme of the relay protection reliability assessment method based on an intelligent oscilloscope are of the same concept. For details not described in detail in the technical scheme of the relay protection reliability assessment system based on an intelligent oscilloscope in this embodiment, please refer to the description of the technical scheme of the relay protection reliability assessment method based on an intelligent oscilloscope.

[0095] This embodiment also provides a relay protection reliability evaluation system based on an intelligent oscilloscope, including:

[0096] A multi-source time series correlation data set construction module is used to build a multi-source data acquisition network at each node of the distribution network, collect multi-source recorded wave data, perform time synchronization marking, and construct a multi-source time series correlation data set;

[0097] Fault feature vector extraction module, used to extract fault feature vectors based on multi-source time series correlation data sets;

[0098] The analysis and modeling module is used to construct a distribution network topology model based on fault feature vectors through a graph neural network and establish the fault propagation relationship between nodes. It also uses a recursive neural network to build a collaborative analysis model for multi-source recorded data.

[0099] The evaluation module is used to input real-time recorded data into the collaborative analysis model, calculate the action timing correlation index and fault propagation impact index, evaluate the degree of selective coordination of the protection device and generate an evaluation report.

[0100] This embodiment further provides an electronic device applicable to the relay protection reliability evaluation method based on an intelligent oscilloscope, including:

[0101] Memory and processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the relay protection reliability assessment method based on the intelligent recorder as proposed in the above embodiment.

[0102] This embodiment further provides a storage medium storing a computer program, which, when executed by a processor, implements the relay protection reliability evaluation method based on an intelligent oscilloscope as proposed in the above embodiment.

[0103] The storage medium proposed in this embodiment and the relay protection reliability evaluation method based on the intelligent recorder proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0104] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. 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 solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for evaluating the reliability of relay protection based on an intelligent oscilloscope, characterized in that: include: Build a multi-source data acquisition network at each node of the distribution network to collect multi-source recorded data and perform time synchronization marking to build a multi-source time series correlation data set; Extract fault feature vectors based on multi-source time series correlation datasets; Based on the fault feature vector, a distribution network topology model is constructed through a graph neural network, and the fault propagation relationship between nodes is established. A multi-source recording data collaborative analysis model is constructed using a recursive neural network. The real-time recorded data is input into the collaborative analysis model to calculate the action timing correlation index and fault propagation impact index, evaluate the degree of selective coordination of the protection device and generate an evaluation report.

2. The method for evaluating relay protection reliability based on an intelligent oscilloscope according to claim 1, wherein: The method of constructing a multi-source data acquisition network at each node of the distribution network, collecting multi-source recorded wave data and performing time synchronization marking, and constructing a multi-source time series correlation data set includes: Intelligent recorders are installed at each node of the distribution network to collect multi-source recording data including primary side voltage signals, primary side current signals, secondary side protection device action signals, and secondary side protection device status signals.

3. The relay protection reliability evaluation method based on the intelligent oscilloscope according to claim 2, characterized in that: The multi-source data acquisition network is constructed at each node of the distribution network to collect multi-source recorded wave data and perform time synchronization marking, and the multi-source time series correlation data set is constructed, which further includes: Multi-source recorded data are time-synchronized and marked, a unified microsecond-level time reference is established, and a multi-source time series correlation data set is formed.

4. The method for evaluating relay protection reliability based on an intelligent oscilloscope according to claim 3, wherein: Extracting the fault feature vector based on the multi-source time series correlation data set includes: Eliminate noise interference from multi-source time series correlation data sets through wavelet transform; The fundamental component of a fixed number of sampling points within the period is extracted from the denoised voltage signal. The fundamental amplitude, phase angle and distortion rate are obtained by sliding calculation of adjacent data windows to form a voltage feature vector. The current mutation amount, mutation rate and effective value of the de-noised current signal are calculated to obtain time domain features, and the wavelet energy entropy and singular value decomposition are extracted to obtain frequency domain features. The time domain features and frequency domain features are combined to form a current feature vector. Based on the noise-reduced protection signal, an action sequence feature vector including the action type, action time and duration of the protection device is constructed. A state transition matrix is ​​established, and the state transition probability and state duration are extracted to form the protection signal feature vector.

5. The method for evaluating relay protection reliability based on an intelligent oscilloscope according to claim 4, wherein: The method of constructing a distribution network topology model based on fault feature vectors and establishing fault propagation relationships between nodes through a graph neural network and constructing a multi-source waveform data collaborative analysis model using a recursive neural network includes: A distribution network topology model is constructed based on a graph neural network, fault feature vectors are mapped to the distribution network topology model, and the fault propagation relationship between nodes is established; Recursive neural networks are used to process the fault propagation relationship between nodes and a collaborative analysis model for multi-source recorded data is constructed.

6. The method for evaluating relay protection reliability based on an intelligent oscilloscope according to claim 5, wherein: The inputting of real-time recorded wave data into the collaborative analysis model, calculating the action sequence correlation index and the fault propagation impact index, evaluating the selectivity coordination degree of the protection device and generating an evaluation report includes: The multi-source recorded data collected in real time is input into the multi-source recorded data collaborative analysis model to calculate the action timing correlation index and fault propagation impact index of the relay protection device.

7. The method for evaluating relay protection reliability based on an intelligent oscilloscope according to claim 6, wherein: The inputting of real-time recorded wave data into the collaborative analysis model, calculating the action timing correlation index and the fault propagation impact index, evaluating the selectivity coordination degree of the protection device and generating an evaluation report also includes: Evaluate the selective coordination degree of relay protection devices based on action sequence correlation indicators and fault propagation impact indicators; Generate a coordinated evaluation report for relay protection devices, including evaluation results of coordination timing, fault isolation range, and protection blocking relationships between protection devices.

8. A relay protection reliability evaluation system based on an intelligent oscilloscope, applying the method according to any one of claims 1 to 7, characterized in that: include: A multi-source time series correlation data set construction module is used to build a multi-source data acquisition network at each node of the distribution network, collect multi-source recorded data, perform time synchronization marking, and construct a multi-source time series correlation data set; Fault feature vector extraction module, used to extract fault feature vectors based on multi-source time series correlation data sets; The analysis and modeling module is used to construct a distribution network topology model based on fault feature vectors through a graph neural network and establish the fault propagation relationship between nodes. It also uses a recursive neural network to build a collaborative analysis model for multi-source recorded data. The evaluation module is used to input real-time recorded data into the collaborative analysis model, calculate the action timing correlation index and fault propagation impact index, evaluate the degree of selective coordination of the protection device and generate an evaluation report.

9. An electronic device comprising: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, implement the steps of the method according to any one of claims 1 to 7.

Images