Traffic fault node detection method and device based on character convolutional neural network

Abstract

The invention provides a traffic fault node detection method and device based on a character convolutional neural network. The method comprises the following steps: firstly, obtaining operation data of a target power distribution network and traffic information of each node; establishing a communication state diagram according to the flow information; extracting features of the communication state diagram according to the first neural network model, and converting the features into a character sequence; and finally, recognizing the character sequence according to the character convolutional neural network, and determining a fault node. According to the method, only the graph neural network is utilized to extract the features of the communication state graph, then the features are translated into the character sequence, and whether the node flow is abnormal or not is detected through the efficient and lightweight character convolutional neural network, so that the detection efficiency can be greatly improved while accurate detection is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system safety, and in particular relates to a flow fault node detection method and device based on character convolutional neural network. Background Art

[0002] As network attacks become increasingly sophisticated, power communication networks face threats from hackers, viruses, and other malicious behaviors. Traffic anomaly detection is an important means of identifying these threats. The power communication network is an important part of the power grid control system. Any anomaly may lead to instability or even accidents in the power grid. The power system needs real-time monitoring to ensure timely response and handling of any problems that may affect the operation of the system. By detecting network traffic anomalies, potential problems can be discovered in advance, so that preventive maintenance can be carried out to avoid greater losses.

[0003] Currently, network traffic detection can be achieved through machine learning-based methods. The most commonly used one is graph neural network. However, the model of conventional graph neural network is large and the processing speed is slow. Especially for large power grids with many nodes and complex structures, the communication status graph they generate is more complex and the detection efficiency is lower. Summary of the invention

[0004] In view of this, the present invention provides a traffic fault node detection method and device based on character convolutional neural network, aiming to solve the problem of network traffic detection efficiency in the prior art.

[0005] A first aspect of an embodiment of the present invention provides a method for detecting a traffic fault node based on a character convolutional neural network, comprising:

[0006] Obtain the operating data of the target distribution network and the flow information of each node;

[0007] Based on the traffic information, a communication status diagram is established;

[0008] Extracting features of the communication state graph according to the first neural network model and converting the features into a character sequence;

[0009] The character sequence is recognized based on the character convolutional neural network to determine the faulty node.

[0010] In a possible implementation, a communication state diagram is established according to the traffic information, including:

[0011] Extract node information, edge information and flow information from operation data;

[0012] Determine the topological adjacency matrix based on the node information, edge information and flow information;

[0013] The traffic information of each node is added as an attribute value to each node of the topological adjacency matrix to obtain a communication state graph.

[0014] In a possible implementation, determining a topological adjacency matrix according to node information, edge information, and power flow information includes:

[0015] According to the node information, establish the initial adjacency matrix;

[0016] According to the edge information and flow information, the above initial adjacency matrix is ​​filled to obtain the topological adjacency matrix.

[0017] In a possible implementation, the feature is converted into a character sequence, including:

[0018] The features of the communication state graph are converted into a character sequence according to a preset mapping table.

[0019] In a possible implementation, the feature is converted into a character sequence, including:

[0020] The features of the communication state graph are converted into a character sequence according to a preset hash function.

[0021] In a possible implementation, the feature is converted into a character sequence, including:

[0022] The features of the communication state graph are converted into character sequences according to the pre-built bag-of-words model.

[0023] In a possible implementation, character sequences are identified according to a character convolutional neural network to determine a faulty node, including:

[0024] Input the character sequence into the character convolutional neural network to determine the failure probability of each node;

[0025] The faulty nodes are determined based on the failure probability of each node and the preset threshold.

[0026] A second aspect of an embodiment of the present invention provides a flow fault node detection device based on a character convolutional neural network, comprising:

[0027] An acquisition module is used to obtain the operating data of the target distribution network and the flow information of each node;

[0028] Establishing a module for establishing a communication status diagram based on traffic information;

[0029] A conversion module, used for extracting features of the communication state graph according to the first neural network model, and converting the features into a character sequence;

[0030] The determination module is used to identify the character sequence according to the character convolutional neural network and determine the fault node.

[0031] A third aspect of an embodiment of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the traffic fault node detection method based on a character convolutional neural network of the first aspect are implemented.

[0032] The fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the traffic fault node detection method based on character convolutional neural network as described in the first aspect above.

[0033] The method and device for detecting flow fault nodes based on character convolutional neural networks provided by the embodiments of the present invention first obtain the operating data of the target distribution network and the flow information of each node; then establish a communication status graph based on the flow information; then extract the features of the communication status graph based on the first neural network model, and convert the features into a character sequence; finally, identify the character sequence based on the character convolutional neural network to determine the faulty node. The present invention only uses a graph neural network to extract the features of the communication status graph, and then translates the features into a character sequence. The detection of whether the node flow is abnormal is realized through an efficient and lightweight character convolutional neural network, which can greatly improve the detection efficiency while ensuring accurate detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0035] Figure 1 It is a flow chart of an implementation method of a traffic fault node detection method based on a character convolutional neural network provided by an embodiment of the present invention;

[0036] Figure 2 It is a structural schematic diagram of a flow fault node detection method and device based on a character convolutional neural network provided by an embodiment of the present invention;

[0037] Figure 3 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0038] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0039] Figure 1 is a flow chart of the implementation of the method for detecting traffic fault nodes based on character convolutional neural network provided by an embodiment of the present invention. Figure 1 As shown, in some embodiments, a method for detecting a traffic fault node based on a character convolutional neural network includes:

[0040] S110, obtaining operation data of the target distribution network and flow information of each node;

[0041] S120, establishing a communication status diagram according to the traffic information;

[0042] S130, extracting features of the communication state graph according to the first neural network model, and converting the features into a character sequence;

[0043] S140, identifying the character sequence according to the character convolutional neural network to determine the faulty node.

[0044] In an embodiment of the present invention, if the distribution network is small in scale and is more sensitive to cost, a SCADA system that supports the DNP3.0 communication protocol may be preferred because the protocol is relatively simple and low in cost. For larger-scale distribution networks, the IEC61850 communication protocol has better scalability and interoperability, and can better meet the needs of complex systems. Select remote terminal units and feeder terminal units with high reliability and stability to ensure that they can work normally in various harsh environments. For example, select equipment with characteristics such as waterproof, dustproof, and anti-electromagnetic interference. Check the communication interface of the selected equipment to ensure compatibility with the existing communication network in the distribution network so as to realize remote transmission of data. After the preparation is completed, the sensor can be installed.

[0045] Specifically, at the busbar of the distribution network, select a suitable installation location to ensure that the current transformer and voltage transformer can accurately measure the current and voltage. For example, install sensors at the incoming and outgoing positions of the busbar to obtain comprehensive current and voltage information. For feeders, install sensors at key nodes based on the length of the feeder and the load distribution. For example, install current transformers and voltage transformers at branch points of feeders, important load access points, and other locations.

[0046] Transformers are key equipment in the distribution network. Sensors are installed on the high-voltage and low-voltage sides of the transformer to monitor the operating status of the transformer. To ensure that the sensors are compatible with the data interfaces of the remote terminal unit and the feeder terminal unit, this can be achieved in the following ways:

[0047] Before installing the sensor, carefully read the technical manuals of the remote terminal unit and feeder terminal unit to understand the sensor types and interface specifications they support. Select a sensor that matches the interface of the remote terminal unit and feeder terminal unit, or use an adapter to convert the sensor's output signal into a signal format acceptable to the remote terminal unit and feeder terminal unit.

[0048] Install a GPS receiver at each data collection device to ensure accurate GPS signals are received. Choose a suitable GPS antenna installation location based on the layout and geographical environment of the distribution network to ensure signal reception quality.

[0049] For devices that support the Network Time Protocol, configure them accordingly and set them to obtain time information from a reliable time server. Ensure that the time synchronization accuracy of all data acquisition devices meets the requirements. Generally speaking, the time synchronization error should be less than 1 millisecond. Use the time synchronization monitoring tool to regularly check the time synchronization status of each data acquisition device. If time synchronization anomalies are found, troubleshoot and repair them in a timely manner.

[0050] The traffic at each node of the power grid is usually generated by the following tasks: telemetry data (real-time data collection such as voltage, current, power, etc.), telesignaling data (signal data such as switch status), and remote control instructions (control instructions for switches and other equipment).

[0051] Study the data volume and transmission frequency characteristics of different business types. For example, telemetry data may be sent every few seconds, with a data volume of several bytes each time; telesignaling data is sent when it changes, with a relatively small data volume.

[0052] For each service type, the traffic of a single node is calculated based on its sending frequency and data volume.

[0053] For example, if the telemetry service is sent once every 5 seconds and the data volume is 100 bytes each time, the traffic generated by the service in one second is 100 / 5=20 bytes / second.

[0054] The traffic of all business types on the node is accumulated to obtain the total communication traffic of the node.

[0055] The above method can be used to collect the grid operation data and flow. After the data is collected, preprocessing is required.

[0056] In some embodiments, a communication status graph is established based on traffic information, including: extracting node information, edge information and flow information in operating data; determining a topological adjacency matrix based on the node information, edge information and flow information; adding the traffic information of each node as an attribute value to each node of the topological adjacency matrix to obtain a communication status graph.

[0057] In the embodiment of the present invention, in the preparation stage of the solution implementation, it is necessary to define nodes and edges. The node definition is as follows:

[0058] Assign a unique identifier to each device, which can be a number or a combination of letters and numbers. For example, a bus can be represented by "BM_1", a line can be represented by "LINE_2", and so on. Specify the type of device, including bus, line, transformer, etc. You can use enumeration types or strings to represent device types for easy distinction and management. Set the status of the device to normal or faulty. You can use Boolean values ​​(True for normal, False for faulty) or specific status codes to represent the device status. For example, 0 for normal and 1 for faulty.

[0059] The edge definition is as follows:

[0060] Determine the type of physical connection between nodes, such as overhead lines, cables, etc. You can use a string to represent the connection type for easy identification and analysis. Measure the physical length of the connection, which can be in meters or kilometers. Ensure the accuracy of the length data so that the impact of line length on parameters such as impedance can be considered in subsequent analysis. Calculate the impedance value of the connection, which can be measured or estimated based on parameters such as line type and length. Impedance values ​​are very important for power flow analysis and fault diagnosis.

[0061] Flow information mainly describes the size and direction of a certain quantity flowing in the network. In the power system, flow usually refers to the flow of power, including active power and reactive power; in other fields, flow can also refer to the flow of goods, the amount of information transmitted, etc.

[0062] Flow information is usually associated with edges, that is, each edge may have a specific flow value. For example, in a power grid, an edge connecting node A and node B may have a certain amount of power flow from A to B.

[0063] In addition, the data graph structure needs to be selected according to the scale of the power grid:

[0064] Sparse graph: If the topology of the distribution network is relatively sparse, that is, there are few connections between nodes, you can choose to use the adjacency list data structure. The adjacency list can effectively store sparse graphs, occupy less storage space, and is convenient when adding and deleting nodes and edges.

[0065] Dense graph: If the topology of the distribution network is dense, that is, there are many connections between nodes, you can choose to use the adjacency matrix data structure. The adjacency matrix can quickly query the connection relationship between nodes, but it occupies a large storage space.

[0066] After completing the above work, the graph can be constructed: in some embodiments, the topological adjacency matrix is ​​determined based on the node information, edge information and flow information, including: establishing an initial adjacency matrix based on the node information; filling the above initial adjacency matrix based on the edge information and flow information to obtain a topological adjacency matrix.

[0067] In an embodiment of the present invention, it is first necessary to determine the total number of nodes in the network based on known node information. The value of N can be determined by counting the range of node numbers or the length of the node list. Then, an N×N two-dimensional matrix A is created, and all elements in the matrix are initialized to 0, indicating that in the initial state, it is assumed that there is no connection between nodes or the weight of the connection is 0. The selection of this initial value can be adjusted according to the needs of the specific problem. For example, if the problem indicates that the situation of no connection should be represented by infinity, then the initial value can be set to a very large value.

[0068] Usually the edge information is stored in some data structure, such as a list, dictionary, etc. If the edge information is stored in a list, each edge can be processed by looping through the list. The two connected node numbers i and j and the weight of the edge are extracted from the edge information. After filling the edge information, the initial adjacency matrix becomes an undirected graph. At this time, the matrix elements can be further adjusted according to the direction and size of the flow to form a directed graph as the topological adjacency matrix.

[0069] First, determine the correspondence between power flow information and edges and nodes. For example, in a power grid, power flow information may be associated with a specific edge, and the direction of the power flow is determined by the two nodes connected by the edge.

[0070] Then, adjust the values ​​of the elements in the adjacency matrix according to the direction and size of the flow. For example, if the flow is flowing from node i to node j, and the flow size is P, the value of A[i][j] can be added or multiplied by the parameter related to the flow size to reflect the influence of the flow. The specific adjustment method depends on the requirements of the problem and the nature of the flow information.

[0071] In some embodiments, converting the features into a character sequence includes: converting the features of the communication state graph into a character sequence according to a preset mapping table.

[0072] In an embodiment of the present invention, a mapping table is designed to map each dimension of a feature vector to a character. ASCII code or other encoding methods can be used. The mapping table should be able to effectively represent the information of the feature vector so that different feature vectors can be mapped to different character sequences. For example, the value range of each dimension of the feature vector can be divided into several intervals, each interval corresponding to a character. Or a hash function is used to map the value of the feature vector to a character in a character set. Select a suitable character set to ensure that the size of the character set is large enough to cover all possible feature values. At the same time, the character set should be easy to process and store. Some characters with specific meanings can be selected to represent different dimensions of the feature vector.

[0073] Coding example: Assume that the dimension of the feature vector is 3, and the value range of each dimension is [0,1]. The value range of each dimension can be divided into 10 intervals, each corresponding to a character in the character set {'0', '1', '2', '3', '4', '5', '6', '7', '8', '9'}. For example, if the value of the first dimension of the feature vector is 0.3, the corresponding character is '3'; if the value of the second dimension is 0.7, the corresponding character is '7'; if the value of the third dimension is 0.1, the corresponding character is '1'. In this way, the feature vector (0.3, 0.7, 0.1) can be mapped to the character sequence '371'. The above is only an example of the present invention, and the actual feature vector and the mapping between the feature vector and the character may be more complicated.

[0074] According to the mapping table, the feature vector is converted into a fixed-length character sequence. For each feature vector, each dimension is traversed and the value of the dimension is mapped to a character according to the mapping table. The characters of all dimensions are combined to form a fixed-length character sequence. If the dimension of the feature vector is less than the fixed length, padding characters (such as spaces or specific symbols) can be used to fill the sequence to a fixed length. If the dimension of the feature vector is greater than the fixed length, it can be truncated or other strategies can be used to handle it.

[0075] For example, assuming the dimension of the feature vector is 5, the mapping table maps the value of each dimension to a character, and the length of the generated character sequence is 8. If the feature vector is (0.2, 0.5, 0.8, 0.1, 0.4), according to the mapping table, it is mapped to the characters '2', '5', '8', '1', '4' respectively. Since the dimension of the feature vector is less than the fixed length, a padding character is needed to fill the sequence. The padding character can be selected as '0', and the generated character sequence is '25814000'.

[0076] In some embodiments, converting the features into a character sequence includes: converting the features of the communication state graph into a character sequence according to a preset hash function.

[0077] In the embodiment of the present invention, a suitable hash function is selected to map each dimension of the feature vector to a hash value of a fixed length. The hash function can be a traditional hash algorithm, such as MD5, SHA-1, etc., or a hash function specially designed for feature conversion.

[0078] For each dimension of the feature vector, compute its hash value using the selected hash function. The hash value can be an integer, string, or other data type, depending on the output of the hash function.

[0079] Convert the hash value of each dimension into a character sequence, using ASCII code or other encoding methods. For example, if the hash value is an integer, you can convert it into a string and then convert each character into the character corresponding to the ASCII code.

[0080] In some embodiments, converting the features into a character sequence includes: converting the features of the communication state graph into a character sequence according to a pre-built bag-of-words model.

[0081] First, extract different feature values ​​from all possible feature vectors and build a vocabulary, namely the bag-of-words model. The vocabulary can be a list or a set, where each element represents a different feature value. For each feature vector, compare the value of each dimension with the elements in the vocabulary to determine its position in the vocabulary. Then, use this position information to encode the feature vector. For example, if there are 1000 different feature values ​​in the vocabulary, a 1000-bit binary vector can be used to represent a feature vector, where each bit corresponds to an element in the vocabulary. If the value of a certain dimension of the feature vector is the 512th element in the vocabulary, then the 512th bit of the corresponding binary vector is 1, and the other bits are 0.

[0082] The encoded feature vector is converted into a character sequence, and binary encoding or other encoding methods can be used. For example, the binary vector can be converted into a hexadecimal string, and then each character can be converted into the character corresponding to the ASCII code.

[0083] The above conversion methods can be selected according to specific application scenarios and requirements. For example, the mapping method is simple and intuitive, but may require a large amount of mapping table storage space; the binary encoding method can represent a larger range of values, but may require a longer sequence length; the hash encoding method can quickly calculate the hash value, but there may be hash conflicts; the bag-of-words model method can process high-dimensional feature vectors, but may lose some feature information. In some cases, more than two conversion methods can be used to convert feature vectors, and the union of the elements of the converted character sequence is used as the character sequence actually used.

[0084] In some embodiments, character sequences are identified based on a character convolutional neural network to determine faulty nodes, including: inputting the character sequence into the character convolutional neural network to determine the fault probability of each node; and determining the faulty node based on the fault probability of each node and a preset threshold.

[0085] The more convolution kernels there are, the more different local features can be extracted; the convolution kernel size determines the field of view of the extracted features; the step size controls the moving step size of the convolution operation on the character sequence. For example, multiple convolution kernels of different sizes can be set, such as 3x1, 4x1, 5x1, etc., to extract local features of different lengths. The convolution operation can be implemented using the convolution function in common deep learning frameworks (such as TensorFlow, PyTorch). Max pooling is a commonly used pooling method that selects the maximum value in the local area on each feature map as the output. In general, the pooling window size can be associated with the convolution kernel size, and the step size can be set to the same or smaller value as the pooling window size. Add multiple fully connected layers, each of which can use a different number of neurons. Nonlinear activation functions such as ReLU, Sigmoid, etc. can be used between fully connected layers to perform nonlinear combinations of features. The output dimension of the last fully connected layer is usually the number of nodes in the distribution network, indicating the failure probability of each node.

[0086] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0087] Figure 2 Schematic diagram of the structure of a flow fault node detection method and device based on a character convolutional neural network provided by an embodiment of the present invention. Figure 2 As shown, in some embodiments, a flow fault node detection method and device 2 based on a character convolutional neural network includes:

[0088] An acquisition module 210 is used to acquire the operation data of the target distribution network and the flow information of each node;

[0089] Establishing module 220, used to establish a communication status diagram according to the flow information;

[0090] A conversion module 230, configured to extract features of the communication state graph according to the first neural network model, and convert the features into a character sequence;

[0091] The determination module 240 is used to identify the character sequence according to the character convolutional neural network and determine the faulty node.

[0092] Optionally, establish module 220 to extract node information, edge information and flow information from the operating data; determine the topological adjacency matrix based on the node information, edge information and flow information; add the flow information of each node as an attribute value to each node of the topological adjacency matrix to obtain a communication status diagram.

[0093] Optionally, a module 220 is established to establish an initial adjacency matrix according to the node information; and fill the initial adjacency matrix according to the edge information and the flow information to obtain a topological adjacency matrix.

[0094] Optionally, the conversion module 230 is used to convert the characteristics of the communication state diagram into a character sequence according to a preset mapping table.

[0095] Optionally, the conversion module 230 is used to convert the features of the communication state graph into a character sequence according to a preset hash function.

[0096] Optionally, the conversion module 230 is used to convert the features of the communication state graph into a character sequence according to a pre-built bag-of-words model.

[0097] Optionally, the determination module 240 is used to input the character sequence into the character convolutional neural network to determine the failure probability of each node; and determine the faulty node based on the failure probability of each node and a preset threshold.

[0098] The flow fault node detection method and device based on character convolutional neural network provided in this embodiment can be used to execute the above method embodiments. Its implementation principle and technical effects are similar, and this embodiment will not be repeated here.

[0099] Figure 3 Schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 3 As shown, an electronic device 3 provided by an embodiment of the present invention includes: a processor 30, a memory 31, and a computer program 32 stored in the memory 31 and executable on the processor 30. When the processor 30 executes the computer program 32, the steps in the above-mentioned output voltage calibration method based on LLC are implemented, for example Figure 2 Alternatively, when the processor 30 executes the computer program 32, the functions of each module / unit in the above-mentioned system embodiments are implemented, for example Figure 3 The functions of each module are shown.

[0100] Exemplarily, the computer program 32 may be divided into one or more modules / units, one or more modules / units are stored in the memory 31, and are executed by the processor 30 to implement the present invention. One or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program 32 in the electronic device 3.

[0101] The electronic device 3 may be a mobile phone, MCU, ECU, industrial computer, etc., which is not limited here. The electronic device 3 may include, but is not limited to, a processor 30 and a memory 31. Those skilled in the art will understand that Figure 3 It is only an example of the electronic device 3 and does not constitute a limitation of the electronic device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0102] The processor 30 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0103] The memory 31 may be an internal storage unit of the electronic device 3, such as a hard disk or memory of the electronic device 3. The memory 31 may also be an external storage device of the electronic device 3, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 3. Further, the memory 31 may also include both an internal storage unit of the electronic device 3 and an external storage device. The memory 31 is used to store computer programs and other programs and data required by the electronic device. The memory 31 may also be used to temporarily store data that has been output or is to be output.

[0104] An embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps in the above-mentioned LLC-based output voltage calibration method embodiment are implemented.

[0105] The computer readable storage medium stores a computer program 32, which includes program instructions. When the program instructions are executed by the processor 30, all or part of the processes in the above-mentioned embodiment method are implemented. The computer program 32 can also be used to instruct related hardware to complete. The computer program 32 can be stored in a computer readable storage medium. When the computer program 32 is executed by the processor 30, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program 32 includes computer program code, which can be in source code form, object code form, executable file or some intermediate form. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium.

[0106] The computer-readable storage medium may be an internal storage unit of the electronic device of any of the aforementioned embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Further, the computer-readable storage medium may also include both an internal storage unit of the electronic device and an external storage device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0107] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0108] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0109] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0110] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0111] In the embodiments provided by the present invention, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0112] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0113] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0114] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable media may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0115] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A flow fault node detection method based on character convolutional neural network, characterized in that: include: Obtain the operating data of the target distribution network and the flow information of each node; Establishing a communication status diagram according to the flow information; Extracting features of the communication state graph according to the first neural network model, and converting the features into a character sequence; The character sequence is identified according to the character convolutional neural network to determine the faulty node.

2. The method for detecting traffic fault nodes based on character convolutional neural network according to claim 1 is characterized in that: According to the flow information, a communication state diagram is established, including: Extracting node information, edge information and power flow information from the operation data; Determine a topological adjacency matrix according to the node information, the edge information and the power flow information; The flow information of each node is added as an attribute value to each node of the topological adjacency matrix to obtain the communication state graph.

3. The method for detecting traffic fault nodes based on character convolutional neural network according to claim 2 is characterized in that: Determining a topological adjacency matrix according to the node information, the edge information, and the power flow information includes: According to the node information, an initial adjacency matrix is ​​established; The initial adjacency matrix is ​​filled according to the edge information and the flow information to obtain the topological adjacency matrix.

4. The method for detecting traffic fault nodes based on character convolutional neural network according to claim 1 is characterized in that: Convert the features into a character sequence, including: The features of the communication state diagram are converted into a character sequence according to a preset mapping table.

5. The method for detecting traffic fault nodes based on character convolutional neural network according to claim 1 is characterized in that: Convert the features into a character sequence, including: The features of the communication state graph are converted into a character sequence according to a preset hash function.

6. The method for detecting traffic fault nodes based on character convolutional neural network according to claim 1 is characterized in that: Convert the features into a character sequence, including: The features of the communication state graph are converted into a character sequence according to a pre-built bag-of-words model.

7. The method for detecting traffic fault nodes based on character convolutional neural network according to claim 2 is characterized in that: Recognizing the character sequence according to the character convolutional neural network to determine the fault node includes: Inputting the character sequence into the character convolutional neural network to determine the failure probability of each node; The faulty nodes are determined based on the failure probability of each node and the preset threshold.

8. A flow fault node detection device based on character convolutional neural network, characterized in that: include: An acquisition module is used to obtain the operating data of the target distribution network and the flow information of each node; An establishing module, used for establishing a communication status diagram according to the flow information; A conversion module, used for extracting features of the communication state graph according to the first neural network model, and converting the features into a character sequence; The determination module is used to identify the character sequence according to the character convolutional neural network to determine the fault node.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the traffic fault node detection method based on character convolutional neural network as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the traffic fault node detection method based on a character convolutional neural network as described in any one of claims 1 to 7 above are implemented.

Images