Power equipment monitoring method and system based on big data analysis

Abstract

The invention relates to a power equipment monitoring method and system based on big data analysis, and belongs to the technical field of power monitoring, and the method comprises the steps: an edge end obtains a first physical connection relation between corresponding power equipment and historical monitoring data of the power equipment, obtaining a historical edge feature map; the cloud end obtains a second physical connection relation between the edge ends and historical power transmission data of the edge ends to construct a historical feature map, so that a graph neural network model is constructed; then obtaining a real-time feature map based on the real-time edge feature map and the real-time power transmission data output by each edge end; and the cloud end inputs the real-time feature graph into a graph neural network model, and outputs a monitoring result of the power equipment. According to the invention, the state of the power equipment is monitored so as to analyze the potential abnormity of the power equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power monitoring, and in particular to a method and system for monitoring electric power equipment based on big data analysis. Background Art

[0002] The power system is a unified whole composed of various power equipment such as power generation, power supply (transmission, transformation, distribution), power consumption facilities, and regulation, control and relay protection devices required to ensure its normal operation. The stability of power equipment operation is the basis for the normal operation of the power system, so it is necessary to monitor the power equipment.

[0003] Because power equipment operates under complex and changeable load conditions, its key parameters such as temperature, voltage, and current will show significant differences and dynamic changes, making it difficult for traditional state monitoring methods based on fixed thresholds to accurately determine whether the power equipment is in normal working condition. Not only may it lead to missed detection of hidden faults, but it may also accelerate aging and damage due to long-term operation of the equipment in abnormal conditions, leading to power outages or more serious power accidents. Therefore, how to monitor the status of power equipment to analyze potential anomalies in power equipment is an urgent problem to be solved. Summary of the invention

[0004] In order to monitor the status of power equipment and analyze potential anomalies in the power equipment, the present application provides a power equipment monitoring method and system based on big data analysis.

[0005] In the first aspect, the present application provides a method for monitoring power equipment based on big data analysis, which adopts the following technical solution: A power equipment monitoring method based on big data analysis is applied to a power monitoring system, wherein the power monitoring system includes an edge terminal and a cloud terminal, and the method includes: The edge end obtains a first physical connection relationship between the respective corresponding power devices and historical monitoring data of the power devices; The edge end constructs an edge graph framework based on the first physical connection relationship and the historical monitoring data, adds the historical monitoring data as a node feature to the edge graph framework to obtain a historical edge feature graph, and sends the historical edge feature graph to the cloud; The cloud obtains a second physical connection relationship between each of the edge terminals and historical power transmission data of the edge terminals; The cloud constructs a historical feature map based on the second physical connection relationship, the historical power transmission data, and the historical edge feature map; The cloud constructs a graph neural network model based on the historical feature graph and the operating state of the power equipment under the state of the historical feature graph; The cloud obtains a real-time feature map based on the real-time edge feature map output by each edge terminal and the real-time power transmission data; The cloud inputs the real-time feature graph into the graph neural network model, aggregates and updates the features of each node through the graph neural network model to generate a probability distribution of the state category of each power device, and selects the state category with the highest probability as the monitoring result corresponding to the power device.

[0006] Optionally, the edge end constructs an edge graph framework based on the first physical connection relationship and the historical monitoring data, including: Selecting two of the power devices that do not have a first physical connection relationship with each other from all the power devices corresponding to the edge end as a group of target power devices; Performing correlation analysis on each item of the historical monitoring data of each group of the target power equipment to obtain a correlation coefficient between each group of the target power equipment; Establishing a correlation connection relationship between each group of target power devices based on the correlation coefficient; The edge end constructs an edge graph framework based on the correlation connection relationship and the first physical connection relationship.

[0007] Optionally, performing correlation analysis on each item of the historical monitoring data of each group of the target power equipment to obtain a correlation coefficient between each group of the target power equipment includes: Performing a correlation analysis on each item of the historical monitoring data of the target power equipment and each item of the historical monitoring data of the target power equipment in the same group to obtain a single coefficient; All the single coefficients are superimposed to obtain the correlation coefficient between each group of target power equipment.

[0008] Optionally, the edge end constructs an edge graph framework based on the correlation connection relationship and the first physical connection relationship, including: The edge end regards each of the corresponding power devices as a single node; Establishing an undirected edge based on the correlation connection relationship; Establishing a first directed edge based on the first physical connection relationship; An edge graph framework is constructed based on the single node, the undirected edge and the first directed edge.

[0009] Optionally, the cloud constructs a historical feature map based on the second physical connection relationship, the historical power transmission data, and the historical edge feature map, including: The cloud regards each of the edge terminals as a global node and establishes a second directed edge based on the second physical connection relationship; Building a global graph framework based on the global node and the second directed edge; Adding the historical power transmission data as node features to the global graph framework to obtain a historical global feature graph; Based on the second physical connection relationship, the historical global feature map is combined with each of the historical edge feature maps to obtain a historical feature map.

[0010] Optionally, the cloud constructs a graph neural network model based on the historical feature graph and the operating state of the power equipment under the historical feature graph state, including: The cloud writes the historical feature graph into a training data set as a training sample, and uses the operating status of the power equipment as a training label for each training sample; Input each training sample in the training data set into a preset neural network model to analyze the state of each power device to obtain a probability value of each state category of the power device; wherein the state category includes one or more combinations of normal, overvoltage and overcurrent fault, overtemperature fault, vibration fault and switch fault; The preset neural network model is iteratively updated based on the probability value of the state category and the training label until the iteration termination condition is reached to obtain the graph neural network model.

[0011] Optionally, the cloud obtains a real-time feature map based on the real-time edge feature map and real-time power transmission data output by each edge terminal, including: Adding the real-time power transmission data as node features into the global graph framework to obtain a real-time global feature graph; Based on the second physical connection relationship, the real-time global feature map is combined with each of the real-time edge feature maps to obtain a real-time feature map.

[0012] Optionally, the power equipment includes a transformer, a generator, a circuit breaker, a mutual inductor and a voltage protector; the monitoring data of the power equipment includes transformer monitoring data, generator monitoring data, circuit breaker monitoring data, mutual inductor monitoring data and voltage protector monitoring data; Among them, the transformer monitoring data includes oil temperature, winding temperature, current and voltage values, and vibration frequency; the generator monitoring data includes active and reactive power, current and voltage values, stator temperature, rotor temperature, and bearing temperature; the circuit breaker monitoring data includes switch status and number of operations; the transformer monitoring data includes secondary current value and secondary voltage value; the voltage protector monitoring data includes leakage current and number of actions.

[0013] In the second aspect, the present application provides a power equipment monitoring system based on big data analysis, which adopts the following technical solutions: A power equipment monitoring system based on big data analysis, comprising: an edge terminal and a cloud terminal; The edge end includes: A first state acquisition module, used to acquire a first physical connection relationship between the respective corresponding power devices and historical monitoring data of the power devices; An edge feature generation module, configured to construct an edge graph framework based on the first physical connection relationship and the historical monitoring data, add the historical monitoring data as node features to the edge graph framework to obtain a historical edge feature graph, and send the historical edge feature graph to the cloud; The cloud includes: A second state acquisition module, used to acquire a second physical connection relationship between the edge ends and historical power transmission data of the edge ends; A historical feature generation module, configured to construct a historical feature map based on the second physical connection relationship, the historical power transmission data, and the historical edge feature map; A model building module, used to build a graph neural network model based on the historical feature graph and the operating state of the power equipment under the state of the historical feature graph; A real-time feature generation module, used to obtain a real-time feature map based on the real-time edge feature map output by each edge terminal and the real-time power transmission data; A monitoring module is used to input the real-time feature graph into a graph neural network model, generate a probability distribution of the state category of each of the power equipment after performing feature aggregation and updating on each node through the graph neural network model, and select the state category with the highest probability as the monitoring result corresponding to the power equipment.

[0014] In a third aspect, the present application provides a computer device, which adopts the following technical solution: A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program according to any one of the above methods.

[0015] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium includes a computer program that can be loaded by a processor and executed in any of the above methods.

[0016] In summary, the present application includes the following beneficial technical effects: First, the first physical connection relationship and historical monitoring data between the corresponding power equipment are obtained through the edge end, which together constitute a comprehensive portrait reflecting the operating status of the power equipment, so that the edge end constructs a historical edge feature map based on the historical monitoring data and the first physical connection relationship, and organizes a large amount of data into a historical edge feature map through the edge end to reduce the workload of subsequent cloud computing for data processing. Subsequently, after the cloud receives the historical edge feature map from each edge end, it continues to globally process the historical edge feature map according to the second physical connection relationship and the historical power transmission data, and obtains the historical feature map. In addition to the historical edge feature map information, the historical feature map also incorporates the overall operation of the power grid and the mutual influence between power equipment. Then, based on the historical feature map, the cloud constructs a graph neural network model. When each edge end uploads a new edge feature map and power transmission data in real time, the cloud integrates the edge feature map into the real-time feature map and inputs it into the trained graph neural network model. The graph neural network model then aggregates and updates the features of each node (i.e., each power device) to generate the latest feature representation reflecting the current operating status of the power equipment. Finally, the graph neural network model outputs the probability distribution of the status category of each power device and automatically selects the category with the highest probability as the monitoring result. Through the collaboration between the cloud and the edge, it can not only accurately monitor the operating status of each power device, but also promptly discover potential operating problems or hidden faults. The edge feature map generated by the edge reduces the amount of data processed by the cloud, making the cloud more real-time when monitoring power equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of a method for monitoring electric power equipment based on big data analysis in one embodiment of the present application; Figure 2 This is a flow chart of a method for constructing an edge graph framework in one embodiment of the present application; Figure 3 is a schematic diagram of an edge graph framework in one embodiment of the present application; Figure 4 This is a flow chart of a method for constructing a historical feature map in one embodiment of the present application; Figure 5 is a schematic diagram of a historical characteristic diagram of one embodiment of the present application; Figure 6 This is a detailed flow chart of a method for constructing an edge graph framework in one embodiment of the present application; Figure 7 This is a block diagram of an electric power equipment monitoring system based on big data analysis according to one embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0019] The present application embodiment discloses a method for monitoring power equipment based on big data analysis. Figure 1 , a method for monitoring electric power equipment based on big data analysis, including: applied to an electric power monitoring system, the electric power monitoring system includes an edge terminal and a cloud terminal, the method includes steps S101-step S107, which are described in detail below.

[0020] Step S101: The edge end obtains a first physical connection relationship between respective corresponding power devices and historical monitoring data of the power devices; Among them, each edge end can correspond to a power site, and the power equipment corresponding to the edge end is the power equipment contained in the power site. As an example, the power equipment may include transformers, generators, circuit breakers, mutual inductors, and voltage protectors; the monitoring data of the power equipment includes transformer monitoring data, generator monitoring data, circuit breaker monitoring data, mutual inductor monitoring data, and voltage protector monitoring data; as an example, the transformer monitoring data includes oil temperature, winding temperature, current and voltage values, and vibration frequency; the generator monitoring data includes active and reactive power, current and voltage values, stator temperature, rotor temperature, and bearing temperature; the circuit breaker monitoring data includes switch status and operation times; the mutual inductor monitoring data includes secondary current value and secondary voltage value; the voltage protector monitoring data includes leakage current and operation times.

[0021] Among them, the first physical connection relationship is used to characterize the connection mode between various power equipment through the transmission line. As an example, the first physical connection relationship can be the connection relationship between the transformer and the generator, the generator is directly or through the bus connected to the primary side of the transformer, and the transformer increases the voltage and sends it to the power grid; it can also be the connection relationship between the transformer and the circuit breaker, that is, the transformer and the circuit breaker are usually connected in series to protect the transformer and other power equipment in the event of a fault; it can also be the connection relationship between the mutual inductor and the transformer, that is, the primary side of the mutual inductor is connected to the secondary side of the transformer or the power grid to convert the high voltage or large current of the transformer into low voltage or small current in proportion for use by measuring instruments and protection devices.

[0022] Step S102: The edge end constructs an edge graph framework based on the first physical connection relationship and the historical monitoring data, adds the historical monitoring data as node features to the edge graph framework to obtain a historical edge feature graph, and sends the historical edge feature graph to the cloud; Step S103: The cloud obtains the second physical connection relationship between each edge terminal and the historical power transmission data of each edge terminal; As an example, power transmission data is data closely related to the operating status of power equipment, and the power transmission data may include active power, reactive power, ambient temperature, and voltage fluctuation range. For example, when active power and reactive power increase, the operating temperature of the power equipment will be affected. Although the operating temperature of the power equipment increases, the power equipment is in a normal working state at this time. Therefore, the power transmission data is used so that the model can learn to identify the impact of power output data on the working state of each power equipment.

[0023] Step S104: The cloud constructs a historical feature map based on the second physical connection relationship, the historical power transmission data, and the historical edge feature map; The second physical connection relationship is the connection relationship between the edge ends, that is, the connection relationship established by the transformers in each power site through the transmission line.

[0024] Step S105: The cloud constructs a graph neural network model based on the historical feature graph and the operating status of the power equipment under the historical feature graph; It should be understood that each historical feature graph corresponds to a set of operating states of all power devices at each edge end, that is, the operating state of each power device under the monitoring data represented by the historical feature graph. For example, the operating state can be one or more combinations of normal, overvoltage and overcurrent fault, overtemperature fault, vibration fault and switch fault. Different power devices can be set to the same operating state or different operating states.

[0025] Step S106: The cloud obtains a real-time feature map based on the real-time edge feature map output by each edge terminal and the real-time power transmission data; Step S107: The cloud inputs the real-time feature graph into the graph neural network model, aggregates and updates the features of each node through the graph neural network model to generate a probability distribution of the state category of each power equipment, and selects the state category with the highest probability as the monitoring result of the corresponding power equipment.

[0026] The status category refers to the type of operating status of the power equipment.

[0027] In the above implementation, first, the first physical connection relationship and historical monitoring data between the corresponding power equipment are obtained through the edge end, which together constitute a comprehensive portrait reflecting the operating status of the power equipment, so that the edge end constructs a historical edge feature map based on the historical monitoring data and the first physical connection relationship, and organizes a large amount of data into a historical edge feature map through the edge end to reduce the workload of subsequent cloud computing for data processing. Subsequently, after the cloud receives the historical edge feature map from each edge end, it continues to globally process the historical edge feature map according to the second physical connection relationship and the historical power transmission data, and obtains the historical feature map. In addition to the historical edge feature map information, the historical feature map also incorporates the overall operation of the power grid and the mutual influence between power equipment. Then, based on the historical feature map, the cloud constructs a graph neural network model. When each edge end uploads a new edge feature map and power transmission data in real time, the cloud integrates the edge feature map into the real-time feature map and inputs it into the trained graph neural network model. The graph neural network model then aggregates and updates the features of each node (i.e., each power device) to generate the latest feature representation reflecting the current operating status of the power device. Finally, the graph neural network model outputs the probability distribution of the status category of each power device and automatically selects the category with the highest probability as the monitoring result. Through the collaboration between the cloud and the edge, it can not only accurately monitor the operating status of each power device, but also promptly discover potential operating problems or hidden faults. The edge feature map generated by the edge reduces the amount of data processed by the cloud, making the cloud more real-time when monitoring power equipment.

[0028] Reference Figure 2 As an implementation of step S102, step S102 includes steps S1021 to S1024, which are described in detail below.

[0029] Step S1021: selecting two power devices that do not have a first physical connection relationship with each other from all power devices corresponding to the edge end as a group of target power devices; Step S1022: performing correlation analysis on various historical monitoring data of each group of target power equipment to obtain correlation coefficients between each group of target power equipment; Step S1023: establishing a correlation connection relationship between each group of target power devices based on the correlation coefficient; As an example, a correlation connection relationship may be established between each group of target power devices when the correlation coefficient is greater than a preset value; otherwise, it is determined that the correlation is low and no correlation connection relationship is established.

[0030] It should be understood that the operating states of two power devices with a first physical connection relationship usually affect each other, but there may also be a correlation in the operating states between power devices without a first physical connection relationship. For example, when a voltage sag or surge occurs somewhere in the power grid, the voltage protector will monitor the voltage change and may trigger a protective action, and the operation of the disconnector may be affected by the state of the voltage protector. Therefore, although there is no direct physical connection between the voltage protector and the disconnector, the operating states of the two may show a certain correlation due to the fluctuation of the power grid voltage. For another example, when a short circuit or overload occurs in the power grid, the transformer will first monitor the abnormal change of the current to trigger the protector to determine the action of the circuit breaker. In this case, the monitoring data of the transformer and the action state of the circuit breaker also show an indirect correlation. Therefore, the correlation of the operating states between power devices that do not have a first physical connection relationship is expressed by establishing a correlation connection relationship.

[0031] Step S1024: The edge end constructs an edge graph framework based on the correlation connection relationship and the first physical connection relationship.

[0032] As an example, the edge graph framework integrates the correlation connection relationship and the first physical connection relationship at the same time, so that the edge graph framework is not limited to the physical connection between each power device, so that the performance expressed by the edge graph framework is better.

[0033] As an implementation of step S1022, step S1022 includes step S10221-step S10222, which are described in detail below.

[0034] Step S10241: performing correlation analysis on each item of historical monitoring data of the target power equipment and each item of historical monitoring data of the target power equipment in the same group to obtain a single coefficient; As an example, each historical monitoring data can be normalized first to unify the measurement standard. For example, the circuit breaker monitoring data includes the switch status and the number of operations; the transformer monitoring data includes the secondary current value and the secondary voltage value. The one-way coefficient includes the switch status-secondary current value coefficient, the number of operations-secondary current value coefficient, the switch status-secondary voltage value coefficient and the number of operations-secondary voltage value coefficient. That is, each historical monitoring data of a target power equipment is traversed and combined with each historical monitoring data of the target power equipment in the same group, and then the one-way coefficients are calculated pairwise.

[0035] Step S10242: superimpose all the individual coefficients to obtain the correlation coefficients between each group of target power equipment.

[0036] As an example, the larger the correlation coefficient is, the stronger the association between the target electric devices in the group is.

[0037] In the above implementation, power devices that do not have a first physical connection relationship with each other are screened out as target analysis objects. Then, correlation analysis is performed on the historical monitoring data of each group of target power devices to quantify the correlation between power devices that do not have a first physical connection relationship, thereby constructing a correlation connection relationship between the target power devices. Finally, the edge end constructs an edge graph framework based on the correlation connection relationship and the first physical connection relationship, so that the edge graph framework can show a more complex and comprehensive interaction relationship in the power system.

[0038] Reference Figure 3 and Figure 5 As an implementation of step S1024, step S1024 includes step S10241-step S10244, which are described in detail below.

[0039] Step S10241: The edge end treats each corresponding power device as a single node; Step S10242: establishing an undirected edge based on the correlation connection relationship; Step S10243: establishing a first directed edge based on the first physical connection relationship; The first directed edge is used to represent the physical connection relationship between the power devices included in each edge end.

[0040] Step S10244: construct an edge graph framework based on the single node, the undirected edge and the first directed edge.

[0041] Combination Figure 3 , taking the edge end including five monomer nodes as an example, namely, monomer node 1, monomer node 2, monomer node 3, monomer node 4 and monomer node 5, wherein monomer node 1 has a first physical connection relationship with monomer nodes 2, 3, and 5 respectively, therefore, a first directed edge is established between monomer node 1 and monomer nodes 2, 3, and 5 respectively, and monomer node 4 and monomer node 3 have a correlation connection relationship, therefore, an undirected edge is established between monomer node 4 and monomer node 3, thereby forming an edge graph framework.

[0042] In the above implementation, undirected edges are established based on correlation connection relationships so as to share implicit information of non-physical connections in the graph neural network. At the same time, the first physical connection relationship is converted into the first directed edge, retaining the real physical connection and the direction of current flow. Finally, the edge graph framework constructed based on monomer nodes, undirected edges and directed edges makes the information carried by the edge graph framework richer.

[0043] Reference Figure 6 As an implementation of step S104, step S104 includes step S1041-step S1044, which are described in detail below.

[0044] Step S1041: the cloud regards each edge end as a global node and establishes a second directed edge based on the second physical connection relationship; As an example, the global node is used to characterize the overall characteristics of the power site corresponding to the edge end, rather than the characteristics of a single power device. Figure 4 , there is a parallel relationship between the global nodes in the edge end and the edge feature graph, and the global nodes do not belong to the nodes in the edge feature graph.

[0045] Step S1042: constructing a global graph framework based on the global node and the second directed edge; Step S1043: adding historical power transmission data as node features to the global graph framework to obtain a historical global feature graph; Step S1044: Based on the second physical connection relationship, the historical global feature map is combined with each historical edge feature map to obtain a historical feature map.

[0046] In the above implementation, the second directed edge established based on the second physical connection relationship, the physical connection path and energy flow direction in the actual transmission line, so as to interact and transmit information based on the second directed edge. Then the historical power transmission data is used as the feature of the global node to obtain a global feature graph. Finally, a more comprehensive historical feature graph is obtained by combining the global feature graph with each historical edge feature graph. The historical feature graph not only includes the overall output status of each power site, but also integrates the local details and historical data of each edge end, so that when monitoring the operating status of each power equipment at the edge end, the influence of the power site load can be deeply combined, thereby making the monitoring of the operating status of the power equipment more accurate.

[0047] As an implementation of step S105, step S105 includes steps S1051 to S1054, which are described in detail below.

[0048] Step S1051: The cloud writes the historical feature graph into the training data set as a training sample, and uses the operating status of the power equipment as a training label for each training sample; Step S1052: input each training sample in the training data set into a preset neural network model to analyze the state of each power device to obtain the probability value of each state category of the power device; wherein the state category includes one or more combinations of normal, overvoltage and overcurrent fault, overtemperature fault, vibration fault and switch fault; As an example, different power devices may be set to different status categories. For example, the status categories of a transformer may include normal, over-temperature fault, and over-voltage and over-current fault. The status categories of a circuit breaker may include normal and switch fault.

[0049] Step S1053: Iteratively update the preset neural network model based on the probability value of the state category and the training label until the iteration termination condition is reached to obtain a graph neural network model.

[0050] Among them, Graph Neural Network (GNN) refers to the general term for algorithms that use neural networks to learn graph structured data, extract and discover features and patterns in graph structured data, and meet the needs of graph learning tasks such as clustering, classification, prediction, segmentation, and generation. By adopting a graph neural model, it can output the status of each power device in each edge end and make full use of a large amount of power data (including monitoring data of each power device and power transmission data).

[0051] As an implementation of step S106, step S106 includes step S1061-step S1062, which are described in detail below.

[0052] Step S1061: adding the real-time power transmission data as node features to the global graph framework to obtain a real-time global feature graph; Step S1062: Based on the second physical connection relationship, the real-time global feature map is combined with each real-time edge feature map to obtain a real-time feature map.

[0053] It should be understood that steps S1061 and S1062 are for generating real-time feature maps in the monitoring phase, and the principle thereof is consistent with the principle of generating historical feature maps in the training phase in steps S1041 and S1042, and will not be repeated here.

[0054] In addition, the embodiment of the present application discloses a power equipment monitoring system based on big data analysis. The power equipment monitoring system based on big data analysis can be applied to a computer device, which is a schematic diagram of the architecture of a computer device for implementing the above method provided by an embodiment of the present invention. In this embodiment, the computer device may include a power equipment monitoring system based on big data analysis, a machine-readable storage medium, and a processor.

[0055] In this embodiment, the machine-readable storage medium and the processor may be located in the computer device and the two are separately arranged. The machine-readable storage medium may also be independent of the computer device and accessed by the processor. The power equipment monitoring system based on big data analysis may include multiple functional modules stored in the machine-readable storage medium, such as the various software functional modules included in the power equipment monitoring system based on big data analysis. When the processor executes the computer program corresponding to the software functional module in the power equipment monitoring system based on big data analysis, the power equipment monitoring system based on big data analysis provided by the aforementioned method embodiment is implemented.

[0056] In this embodiment, the computer device may include one or more processors. The processor may process information and / or data related to the service request to perform one or more functions described in the present invention. In some embodiments, the processor may include one or more processing engines (e.g., a single-core processor or a multi-core processor). Just to give an example, the processor may include one or more hardware processors, such as a central processing unit CPU, an application-specific integrated circuit ASIC, a dedicated instruction set processor ASIP, a graphics processor GPU, a physical operation processing unit PPU, a digital signal processor DSP, a field programmable gate array FPGA, a programmable logic device PLD, a controller, a microcontroller unit, a reduced instruction set computer RISC, a microprocessor, etc., or similar or any combination thereof.

[0057] The machine-readable storage medium may store data and / or instructions. In some embodiments, the machine-readable storage medium may store acquired data or information. In some embodiments, the machine-readable storage medium may store data and / or instructions for execution or use by the computer device, and the computer device may implement the exemplary method described in the present application by executing or using the data and / or instructions. In some embodiments, the machine-readable storage medium may include a mass storage, a removable storage, a volatile read-write memory, a read-only memory ROM, etc., or any combination of the above examples. Exemplary mass storage may include a disk, an optical disk, a solid-state disk, etc. Exemplary removable storage may include a flash drive, a floppy disk, an optical disk, a memory card, a compressed disk, a magnetic tape, etc. Exemplary volatile read-write memory may include a random access memory RAM. Exemplary random access memory may include dynamic RAM, double rate synchronous dynamic RAM, static RAM, thyristor RAM, and zero capacitance RAM, etc. Exemplary ROM may include a masked ROM, a programmable ROM, an erasable programmable ROM, an electronically erasable programmable ROM, a compressed disk ROM, and a digital universal disk ROM, etc.

[0058] The power equipment monitoring system based on big data analysis included in the computer device may include one or more software function modules. The software function modules may be programs and instructions stored in the machine-readable storage medium, and these software function modules are used to implement the above method when executed by the corresponding processor, for example, when executed by the processor of the drone, they are used to implement the method steps executed by the drone, or when executed by the computer device, they are used to implement the method steps executed by the computer device.

[0059] Refer to the detailed Figure 7, the embodiment of the present application discloses a power equipment monitoring system based on big data analysis. A power equipment monitoring system based on big data analysis includes: an edge terminal and a cloud terminal; The edge end includes: A first state acquisition module, used to acquire a first physical connection relationship between respective corresponding power devices and historical monitoring data of the power devices; An edge feature generation module, configured to construct an edge graph framework based on the first physical connection relationship and the historical monitoring data, add the historical monitoring data as node features to the edge graph framework to obtain a historical edge feature graph, and send the historical edge feature graph to the cloud; The cloud includes: A second state acquisition module is used to acquire a second physical connection relationship between each edge end and historical power transmission data of each edge end; A historical feature generation module, used to construct a historical feature map based on the second physical connection relationship, the historical power transmission data and the historical edge feature map; A model building module, used to build a graph neural network model based on the historical feature graph and the operating status of the power equipment under the historical feature graph; A real-time feature generation module, used to obtain a real-time feature map based on the real-time edge feature map output by each edge terminal and the real-time power transmission data; The monitoring module is used to input the real-time feature graph into the graph neural network model, aggregate and update the features of each node through the graph neural network model to generate the probability distribution of the state category of each power equipment, and select the state category with the highest probability as the monitoring result of the corresponding power equipment.

[0060] The electric power equipment monitoring system based on big data analysis provided in the present application can realize the above-mentioned electric power equipment monitoring method based on big data analysis, and the specific working process of the electric power equipment monitoring system based on big data analysis can refer to the corresponding process in the above-mentioned method embodiment.

[0061] It should be noted that in the above embodiments, the description of each embodiment has different emphases, and for the parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0062] The present invention also discloses a computer-readable storage medium, which includes a computer program that can be loaded by a processor and executed as in any of the above methods.

[0063] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the 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 communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0064] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0065] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. A power equipment monitoring method based on big data analysis, applied to a power monitoring system, the power monitoring system comprising an edge terminal and a cloud terminal, characterized in that: The method comprises: The edge end obtains a first physical connection relationship between the respective corresponding power devices and historical monitoring data of the power devices; The edge end constructs an edge graph framework based on the first physical connection relationship and the historical monitoring data, adds the historical monitoring data as a node feature to the edge graph framework to obtain a historical edge feature graph, and sends the historical edge feature graph to the cloud; The cloud obtains a second physical connection relationship between each of the edge terminals and historical power transmission data of the edge terminals; The cloud constructs a historical feature map based on the second physical connection relationship, the historical power transmission data, and the historical edge feature map; The cloud constructs a graph neural network model based on the historical feature graph and the operating state of the power equipment under the state of the historical feature graph; The cloud obtains a real-time feature map based on the real-time edge feature map output by each edge terminal and the real-time power transmission data; The cloud inputs the real-time feature graph into the graph neural network model, and generates a probability distribution of the state category of each power device after performing feature aggregation and updating on each node through the graph neural network model, and selects the state category with the highest probability as the monitoring result corresponding to the power device.

2. The method according to claim 1, characterized in that: The edge end constructs an edge graph framework based on the first physical connection relationship and the historical monitoring data, including: Selecting two of the power devices that do not have a first physical connection relationship with each other from all the power devices corresponding to the edge end as a group of target power devices; Performing correlation analysis on each item of the historical monitoring data of each group of the target power equipment to obtain a correlation coefficient between each group of the target power equipment; Establishing a correlation connection relationship between each group of target power devices based on the correlation coefficient; The edge end constructs an edge graph framework based on the correlation connection relationship and the first physical connection relationship.

3. The method according to claim 2, characterized in that The performing correlation analysis on each item of the historical monitoring data of each group of the target power equipment to obtain the correlation coefficient between each group of the target power equipment includes: Performing a correlation analysis on each item of the historical monitoring data of the target power equipment and each item of the historical monitoring data of the target power equipment in the same group to obtain a single coefficient; All the single coefficients are superimposed to obtain the correlation coefficient between each group of target power equipment.

4. The method according to claim 2, characterized in that: The edge end constructs an edge graph framework based on the correlation connection relationship and the first physical connection relationship, including: The edge end regards each of the corresponding power devices as a single node; Establishing an undirected edge based on the correlation connection relationship; Establishing a first directed edge based on the first physical connection relationship; An edge graph framework is constructed based on the single node, the undirected edge and the first directed edge.

5. The method according to claim 1, characterized in that The cloud constructs a historical feature map based on the second physical connection relationship, the historical power transmission data, and the historical edge feature map, including: The cloud regards each of the edge terminals as a global node and establishes a second directed edge based on the second physical connection relationship; Building a global graph framework based on the global node and the second directed edge; Adding the historical power transmission data as node features to the global graph framework to obtain a historical global feature graph; Based on the second physical connection relationship, the historical global feature map is combined with each of the historical edge feature maps to obtain a historical feature map.

6. The method according to claim 1, characterized in that The cloud constructs a graph neural network model based on the historical feature graph and the operating state of the power equipment under the historical feature graph state, including: The cloud writes the historical feature graph into a training data set as a training sample, and uses the operating status of the power equipment as a training label for each training sample; Input each training sample in the training data set into a preset neural network model to analyze the state of each power device to obtain a probability value of each state category of the power device; wherein the state category includes one or more combinations of normal, overvoltage and overcurrent fault, overtemperature fault, vibration fault and switch fault; The preset neural network model is iteratively updated based on the probability value of the state category and the training label until the iteration termination condition is reached to obtain the graph neural network model.

7. The method according to claim 5, characterized in that The cloud obtains a real-time feature map based on the real-time edge feature map and real-time power transmission data output by each edge terminal, including: Adding the real-time power transmission data as node features into the global graph framework to obtain a real-time global feature graph; Based on the second physical connection relationship, the real-time global feature map is combined with each of the real-time edge feature maps to obtain a real-time feature map.

8. The method according to any one of claims 1 to 6, characterized in that: The power equipment includes a transformer, a generator, a circuit breaker, a mutual inductor and a voltage protector; the monitoring data of the power equipment includes transformer monitoring data, generator monitoring data, circuit breaker monitoring data, mutual inductor monitoring data and voltage protector monitoring data; Among them, the transformer monitoring data includes oil temperature, winding temperature, current and voltage values, and vibration frequency; the generator monitoring data includes active and reactive power, current and voltage values, stator temperature, rotor temperature, and bearing temperature; the circuit breaker monitoring data includes switch status and number of operations; the transformer monitoring data includes secondary current value and secondary voltage value; the voltage protector monitoring data includes leakage current and number of actions.

9. A power equipment monitoring system based on big data analysis, characterized in that: Including edge and cloud; The edge end includes: A first state acquisition module, used to acquire a first physical connection relationship between the respective corresponding power devices and historical monitoring data of the power devices; An edge feature generation module, configured to construct an edge graph framework based on the first physical connection relationship and the historical monitoring data, add the historical monitoring data as node features to the edge graph framework to obtain a historical edge feature graph, and send the historical edge feature graph to the cloud; The cloud includes: A second state acquisition module, used to acquire a second physical connection relationship between the edge ends and historical power transmission data of the edge ends; A historical feature generation module, configured to construct a historical feature map based on the second physical connection relationship, the historical power transmission data, and the historical edge feature map; A model building module, used to build a graph neural network model based on the historical feature graph and the operating state of the power equipment under the state of the historical feature graph; A real-time feature generation module, used to obtain a real-time feature map based on the real-time edge feature map output by each edge terminal and the real-time power transmission data; A monitoring module is used to input the real-time feature graph into a graph neural network model, generate a probability distribution of the state category of each of the power equipment after performing feature aggregation and updating on each node through the graph neural network model, and select the state category with the highest probability as the monitoring result corresponding to the power equipment.

10. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program according to any one of the methods of claims 1 to 7.

Images