Isolation switch fault prediction method and system based on intelligent perception and data analysis

Abstract

The invention discloses an isolation switch fault prediction method and system based on intelligent perception and data analysis, and the method comprises the steps: obtaining the vibration data of a to-be-tested device, and the to-be-tested device comprises an isolation switch and a function device; constructing a power relation graph corresponding to the to-be-tested equipment, and determining switch nodes and function nodes in the power relation graph; determining first graph representation information based on the graph information of the function nodes on the power relation graph; determining first vibration characterization information of the function node and second vibration characterization information of the switch node based on the vibration data; determining a first correlation degree between the function node and the switch node; determining a function weight based on the first relevancy and the device attribute of the function device, and combining the first graph representation information and the first vibration representation information based on the function weight to obtain first comprehensive representation information; and the fault information of the switch node is predicted based on the first comprehensive representation information and the second vibration representation information, so that the risk of cascading failures can be effectively handled.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric power, and in particular to a disconnecting switch fault prediction method and system based on intelligent sensing and data analysis. BACKGROUND

[0002] In the field of power system operation and maintenance, as a core component for ensuring the safety of equipment and the stability of the system, the running state of the disconnecting switch is directly related to the reliability and safety of the entire power grid. Research on the fault prediction method of the disconnecting switch not only has important significance for preventing large-scale power outages, but also provides key support for the stable operation of the smart grid.

[0003] In related technologies, the fault prediction of the disconnecting switch is mostly focused on the independent monitoring and diagnosis of a single switching device. For example, by analyzing the historical data of different types of disconnecting switches in advance, the fault characteristics of the corresponding type of disconnecting switch are identified, so that when the data of a certain disconnecting switch meets the corresponding fault characteristics, it is predicted that the switch will fail. However, since power failures often trigger a chain reaction, leading to a chain failure of a certain region or a large-scale device (such as a large substation), the prediction results provided by related technologies are not comprehensive enough to effectively deal with the risk of chain failure. SUMMARY

[0004] To solve the above technical problems, the embodiments of the present application provide a disconnecting switch fault prediction method and system based on intelligent sensing and data analysis, which can effectively deal with the risk of chain failure and improve the accuracy of disconnecting switch fault prediction.

[0005] In a first aspect, the embodiments of the present application provide a disconnecting switch fault prediction method based on intelligent sensing and data analysis, comprising: obtaining vibration data of a plurality of to-be-tested devices through intelligent sensors, wherein the plurality of to-be-tested devices include a plurality of disconnecting switches and functional devices between the disconnecting switches; constructing a power relationship graph corresponding to the plurality of to-be-tested devices, determining a switch node corresponding to the disconnecting switch and a functional node corresponding to the functional device in the power relationship graph; determining first graph feature information of the functional node based on graph information of the functional node on the power relationship graph; determining first vibration feature information of the functional node and second vibration feature information of the switch node based on the vibration data; determining a first correlation degree between the functional node and the switch node in the power relationship graph; determining a function weight of the function node based on the first correlation and the device attribute of the function device, and combining the first graph representation information and the first vibration representation information based on the function weight to obtain first comprehensive representation information of the function node; Fault information corresponding to the switch node is predicted based on the first comprehensive representation information and the second vibration representation information.

[0006] Optionally, determining the first correlation between the functional node and the switch node includes: For each functional node, Determine, among the switch nodes, a first switch node that is connected to the functional node; Determining a second switching node directly connected to the functional node from the first switching nodes; A first correlation corresponding to the functional node is determined based on the first switch node and the second switch node.

[0007] Optionally, determining a first correlation corresponding to the functional node based on the first switch node and the second switch node includes: Determine all routes from the functional node to the first switch node; Based on the information of all the lines, the information of the first switch node, and the information of the second switch node, a first correlation corresponding to the functional node is determined.

[0008] Optionally, constructing a power relationship graph corresponding to the plurality of devices under test includes: Obtaining device information of the multiple devices under test; Taking the multiple devices under test as multiple nodes, and determining each edge between the multiple nodes according to electrical connection information between the multiple devices under test; Determine the weight of each edge according to the device information; The power relationship graph is constructed based on the multiple nodes, the edges and the weights of the edges.

[0009] Optionally, determining the weight of each edge according to the device information includes: For each of the edges, Determining different nodes connected by the edge among the multiple nodes, and determining a second correlation of the different nodes based on device information corresponding to each of the different nodes; Based on the second correlation, a weight of the edge is determined.

[0010] Optionally, constructing the power relationship graph based on the multiple nodes, the edges, and the weights of the edges includes: Constructing an initial relationship graph based on the multiple nodes, the edges, and the weights of the edges; Determining an initial functional node corresponding to each functional device in the initial relationship graph; Determining the functional criticality of each initial functional node on the initial relationship graph based on the weights of all edges connected to each initial functional node and the initial relationship graph; Based on the functional criticality of each initial functional node, the weights of all edges connected to each initial functional node are updated to update the initial relationship graph, and the updated initial relationship graph is used as the power relationship graph.

[0011] Optionally, predicting fault information corresponding to the switch node based on the first comprehensive representation information and the second vibration representation information includes: Predicting a failure probability corresponding to each switch node based on the first comprehensive characterization information and the second vibration characterization information; For each switch node, if its corresponding fault probability meets a preset condition, real-time monitoring data corresponding to the switch node is obtained, and fault information corresponding to the switch node is determined based on the real-time monitoring data and the second vibration characterization information.

[0012] Optionally, the real-time monitoring data corresponding to the switch node includes: line voltage fluctuation data and operating status data corresponding to the switch node, data of the isolating switch and adjacent devices corresponding to the switch node, and first environmental data of an area where the isolating switch corresponding to the switch node is located, wherein the adjacent devices are devices adjacent to the isolating switch corresponding to the switch node among the multiple devices under test; The determining, based on the real-time monitoring data and the second vibration characterization information, fault information corresponding to the switch node includes: Determining a device state indicator characteristic of the switch node based on the line voltage fluctuation data and the second vibration characterization information; Determining a potential risk level of early failure based on the equipment status indicator characteristics; When the potential risk level of the early fault is higher than the preset level, the fault information corresponding to the switch node is determined based on the data of the disconnector and adjacent equipment corresponding to the switch node, as well as the first environment data and the operating status data.

[0013] Optionally, the data of the isolating switch and adjacent equipment corresponding to the switch node include: accumulated data of thermal effects of the isolating switch and adjacent equipment corresponding to the switch node, and partial discharge data between the isolating switch and adjacent equipment corresponding to the switch node; The determining, based on the data of the isolating switch and adjacent equipment corresponding to the switch node, as well as the first environment data and the operating status data, the fault information corresponding to the switch node includes: Determining a fault range based on the thermal effect accumulation data and the partial discharge data; Determining second environmental data in the first environmental data that matches the fault range; Fault information corresponding to the switch node is determined based on the second environment data and the operating status data.

[0014] In a second aspect, an embodiment of the present application provides an isolating switch fault prediction system based on intelligent perception and data analysis, including: A vibration data acquisition module, configured to acquire vibration data of a plurality of devices under test through an intelligent sensor, wherein the plurality of devices under test include a plurality of isolating switches and functional devices between the isolating switches; A graph module, configured to construct a power relationship graph corresponding to the plurality of devices under test, and determine in the power relationship graph the switch node corresponding to the isolation switch and the function node corresponding to the function device; A graph representation module, configured to determine first graph representation information of the functional node based on graph information of the functional node on the power relationship graph; a vibration characterization module, configured to determine first vibration characterization information of the functional node and second vibration characterization information of the switch node based on the vibration data; A correlation module, configured to determine, in the power relationship graph, a first correlation between the function node and the switch node; a comprehensive characterization module, configured to determine a function weight of the function node based on the first correlation and the device attribute of the function device, and combine the first graph characterization information and the first vibration characterization information based on the function weight to obtain first comprehensive characterization information of the function node; A prediction module is used to predict fault information corresponding to the switch node based on the first comprehensive characterization information and the second vibration characterization information.

[0015] In summary, the embodiments of the present application have at least the following beneficial effects: According to an embodiment of the present application, vibration data of multiple devices under test are obtained through an intelligent sensor, wherein the multiple devices under test include multiple isolating switches and functional devices between the isolating switches; a power relationship graph corresponding to the multiple devices under test is constructed, and the switch node corresponding to the isolating switch and the functional node corresponding to the functional device are determined in the power relationship graph; based on the graph information of the functional node on the power relationship graph, the first graph representation information of the functional node is determined; based on the vibration data, the first vibration representation information of the functional node and the second vibration representation information of the switch node are determined; in the power relationship graph, the first correlation between the functional node and the switch node is determined; based on the first correlation and the device attributes of the functional device, the functional weight of the functional node is determined, and the first graph representation information and the first vibration representation information are merged based on the functional weight to obtain the first comprehensive representation information of the functional node; based on the first comprehensive representation information and the second vibration representation information, the fault information corresponding to the switch node is predicted, so as to effectively deal with the risk of chain failures and improve the accuracy of isolating switch fault prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of a disconnector fault prediction method based on intelligent perception and data analysis provided in an embodiment of the present application; Figure 2 is a schematic diagram of an isolating switch fault prediction system based on intelligent perception and data analysis provided in an embodiment of the present application; Figure 3 It is a schematic diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0018] In the description of this application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, features defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. In the description of this application, unless otherwise specified, "multiple" means two or more. In the description of this application, the term "including" and its variations are open inclusions, i.e., "including but not limited to". The term "based on" means "at least partially based on". The term "according to" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments".

[0019] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in this application based on the specific circumstances.

[0020] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meanings as those commonly understood by those skilled in the art. The terms used in this specification are only for the purpose of describing specific embodiments and are not intended to limit this application. Those of ordinary skill in the art will understand the specific meanings of the above terms in this application in specific circumstances.

[0021] First, see Figure 1 , shows a flow chart of a disconnector fault prediction method based on intelligent perception and data analysis provided in an embodiment of the present application, the method includes S101-S107, as follows.

[0022] S101 , obtaining vibration data of a plurality of devices under test through an intelligent sensor, wherein the plurality of devices under test include a plurality of isolating switches and functional devices between the isolating switches.

[0023] In some examples, the plurality of devices under test may be installed with the smart sensor, which is used to detect vibration conditions of the plurality of devices under test to generate the vibration data.

[0024] In some examples, the above-mentioned functional devices may refer to devices within the power system used to implement various electrical functions. For example, the above-mentioned functional devices may include at least one of the following: at least one circuit breaker, at least one grounding switch, at least one lightning arrester, at least one current transformer, at least one voltage transformer, at least one load switch, and at least one insulator.

[0025] S102: Construct a power relationship diagram corresponding to the plurality of devices under test, and determine a switch node corresponding to the isolation switch and a function node corresponding to the function device in the power relationship diagram.

[0026] In some examples, the above-mentioned multiple devices under test can be directly used as nodes in the graph, and the electrical connection relationship between the above-mentioned multiple devices under test can be used as the edge of the node in the graph, so as to construct a corresponding power relationship graph. In this way, the node in the graph corresponding to the isolating switch in the above-mentioned multiple devices under test is the above-mentioned switch node, and the node in the graph corresponding to the functional device in the above-mentioned multiple devices under test is the above-mentioned functional node.

[0027] S103: Determine first graph representation information of the function node based on the graph information of the function node on the power relationship graph.

[0028] In some examples, the above-mentioned graph information and / or the above-mentioned first graph representation information can be used to indicate relevant information of the functional node in the power relationship graph. For example, the relevant information may include the weight of at least one of all edges connected to the functional node in the power relationship graph, and / or, the relevant information may also include at least one of the following: the position on the power relationship graph, the electrical connection relationship between the node and other nodes on the power relationship graph. It is not difficult to understand that the other nodes here may include one or more other functional nodes and / or one or more switch nodes.

[0029] In some examples, determining first graph representation information of the function node based on graph information of the function node on the power relationship graph may include: obtaining the first graph representation information of the function node by converting the graph information into a target space. The target space may include any of the following: a low-dimensional space, a high-dimensional space, and a latent space.

[0030] S104 : Determine first vibration characterization information of the functional node and second vibration characterization information of the switch node based on the vibration data.

[0031] In some examples, determining the first vibration characterization information of the functional node and the second vibration characterization information of the switch node based on the vibration data may include: converting the vibration data of the functional device into a target space to obtain the first vibration characterization information of the functional node, and converting the vibration data of the isolation switch into a target space to obtain the second vibration characterization information of the switch node. The target space may include any of the following: a low-dimensional space, a high-dimensional space, and a latent space.

[0032] S105 : Determine a first correlation between the function node and the switch node in the power relationship graph.

[0033] In some examples, determining the first correlation between the functional node and the switch node may include: determining a node connection relationship between the functional node and the switch node in a power relationship graph, so as to determine the first correlation between the functional node and the switch node based on the node connection relationship. The node connection relationship may be used to indicate a connection relationship (e.g., direct connection and / or indirect connection) between each functional node and each switch node, so that the first correlation between each functional node and each switch node may be determined based on information (e.g., quantity) of switch nodes directly and / or indirectly connected to the functional node.

[0034] S106 , determining a functional weight of the functional node based on the first correlation and the device attributes of the functional device, and merging the first graph representation information and the first vibration representation information based on the functional weight to obtain first comprehensive representation information of the functional node.

[0035] In some examples, the first correlation can be used to indicate the compactness of the relationship between the functional node (i.e., the corresponding functional device) and each switch node (i.e., each isolating switch). Thus, the first correlation can be used to indicate whether the correlation between the corresponding functional device and each isolating switch is sufficiently compact. For example, the functional weight can increase as the first correlation increases.

[0036] In some examples, the device attributes may include at least one of the following information about the functional device: device value, device precision, functional importance, current capacity, operating life, maintenance frequency, etc.

[0037] In some examples, the initial function weight, the first mapping relationship and the second mapping relationship can be preset, the first mapping relationship is used to indicate different first correlation degrees and corresponding different first weight increase values, and the second mapping relationship is used to indicate different device attributes and corresponding different second weight increase values. In this way, the first weight increase value corresponding to the first correlation degree and the second weight increase value corresponding to the device attribute of the function device can be added to the initial function weight to obtain the final function weight.

[0038] In some examples, the function weight can be determined according to the first correlation degree and the device attribute by using a pre-trained function weight model. The function weight model can be a model trained to have a prediction ability of taking the first correlation degree and the device attribute as model input and taking the function weight as model output. In specific training, sample first correlation degrees and sample device attributes can be used as sample data (the sample data also carries a corresponding expected weight label, which represents the corresponding expected function weight), and a general training algorithm (for example, gradient descent method) is used to train the model, so that the trained model can have the above-mentioned ability. Further, the first mapping relationship and / or the second mapping relationship can also be generated by the function weight model.

[0039] S107, based on the first comprehensive representation information and the second vibration representation information, predicting the fault information corresponding to the switch node.

[0040] In some examples, the fault information corresponding to the switch node can be predicted only according to the first comprehensive representation information and the second vibration representation information, for example, by using a pre-trained fault prediction model to determine the fault information corresponding to the switch node according to the first comprehensive representation information and the second vibration representation information. The fault prediction model can be a model trained to have a prediction ability of taking the first comprehensive representation information and the second vibration representation information as model input and taking the fault information as model output. In specific training, sample comprehensive representation information and sample second vibration representation information can be used as sample data (the sample data also carries a corresponding expected fault label, which represents the corresponding expected fault information), and a general training algorithm (for example, gradient descent method) is used to train the model, so that the trained model can have the above-mentioned ability.

[0041] It should be noted that in modern power systems, disconnectors and their adjacent functional devices (such as circuit breakers, transformers, and lightning arresters) form a complex electrical connection network. During operation, these devices may fail due to factors such as mechanical wear, insulation aging, and abnormal vibration, thereby affecting the stability and safety of the entire system. To achieve fault prediction and status assessment for disconnectors, relying solely on monitoring data from a single disconnector is often inadequate for accurately assessing potential risks. Therefore, this embodiment introduces a more comprehensive information fusion mechanism to improve the robustness and accuracy of the prediction model. Specifically, this embodiment constructs a power relationship graph, combines the topological relationships between functional nodes and switch nodes, and the corresponding attributes of the functional nodes, calculates the functional weights of the functional nodes, and then weights and fuses their representation information to assist in fault prediction for the switch nodes, thereby improving prediction accuracy. Thus, during fault prediction, for each switch node, the corresponding fault information can be accurately predicted based on the first comprehensive representation information corresponding to each functional node with which it is connected, as well as the second vibration representation information of the switch node itself.

[0042] In an optional implementation, determining the first correlation between the functional node and the switch node includes: For each functional node, Determine, among the switch nodes, a first switch node that is connected to the functional node; Determining a second switching node directly connected to the functional node from the first switching nodes; A first correlation corresponding to the functional node is determined based on the first switch node and the second switch node.

[0043] In some examples, the connection relationship may include direct connection and indirect connection, wherein the indirect connection refers to the connection between two nodes via at least one node.

[0044] In some examples, the first correlation corresponding to the functional node can be determined according to the number of first switch nodes and the number of second switch nodes. For example, the greater the number, the higher the corresponding first correlation.

[0045] It is not difficult to understand that the first switch node is used to represent all the isolating switches that are electrically connected to the functional device corresponding to the functional node. The number of first switch nodes can be used to measure the influence range of the functional device corresponding to the functional node in the power network composed of multiple devices under test. If a functional node is connected to more first switch nodes, it means that its position in the power network is more critical. The more first switch nodes there are, the more the fault corresponding to the functional node will affect a wider system area, so the first correlation is higher.

[0046] It's easy to understand that the second switch node represents the disconnector directly electrically connected to the functional device corresponding to the functional node. The second switch node represents the factor that most directly affects the operating state of the functional device corresponding to the functional node. A direct connection indicates a stronger dependency and / or interaction between the two. A greater number of directly connected switch nodes indicates a more important role for the functional node in the power network, as changes in the state of the functional node will more directly affect more adjacent switch nodes, resulting in a higher first correlation.

[0047] In an optional implementation, determining the first correlation corresponding to the functional node based on the first switch node and the second switch node includes: Determine all routes from the functional node to the first switch node; Based on the information of all the lines, the information of the first switch node, and the information of the second switch node, a first correlation corresponding to the functional node is determined.

[0048] In some examples, the information about all the lines may include at least one of the following: the number of all the lines, and the length of each of the lines.

[0049] In some examples, the information of the first switch node may include at least one of the following: the number of first switch nodes, the number of all functional nodes connected to the first switch node, and device attributes of functional devices corresponding to all functional nodes connected to the first switch node.

[0050] In some examples, the information about the second switch node may include at least one of the following: the number of second switch nodes, the number of all functional nodes connected to the second switch node, and device attributes of functional devices corresponding to all functional nodes connected to the second switch node.

[0051] It should be noted that all the above-mentioned lines refer to all lines from the functional node to each first switch node. For example, if a functional node is directly connected to a first switch node, the corresponding lines may not only have one line directly connecting the two, but may also include one or more indirectly connected lines between the two. The indirectly connected lines will be composed of at least one node, and the at least one node may be composed of a switch node and / or a functional node.

[0052] In an optional implementation, constructing the power relationship graph corresponding to the plurality of devices under test includes: Obtaining device information of the multiple devices under test; Taking the multiple devices under test as multiple nodes, and determining each edge between the multiple nodes according to electrical connection information between the multiple devices under test; Determine the weight of each edge according to the device information; The power relationship graph is constructed based on the multiple nodes, the edges and the weights of the edges.

[0053] In some examples, the device information may include at least one of the following: device location, device attributes (such as device attributes of a functional device and / or switch attributes of an isolation switch).

[0054] In some examples, the multiple nodes and the edges can be directly used as nodes and edges in the graph respectively, and the weight of each edge can be used as the weight of the edge in the graph to construct a power relationship graph.

[0055] In an optional implementation, determining the weight of each edge according to the device information includes: For each of the edges, Determining different nodes connected by the edge among the multiple nodes, and determining a second correlation of the different nodes based on device information corresponding to each of the different nodes; Based on the second correlation, a weight of the edge is determined.

[0056] In some examples, based on the device information corresponding to each of the different nodes, at least one of the following can be determined: the functional synergy of the different devices under test represented by different nodes (such as whether there is a cooperative relationship in operation, protection, measurement and other functions); the spatial proximity of the different devices under test represented by different nodes (such as whether they are installed in the same cabinet, interval or bus section); the parameter consistency of the different devices under test represented by different nodes (such as whether the rated voltage and current levels are consistent or matched); the historical behavior similarity of the different devices under test represented by different nodes (such as whether the change trends of historical vibration, temperature, fault and other data are synchronized).

[0057] Continuing with the above example, the second correlation can be obtained by weighted calculation of functional synergy, spatial proximity, parameter consistency and / or historical behavior similarity.

[0058] In some examples, the higher the second correlation, the higher the weight of the edge. In specific implementations, the weight of the edge can be determined based on the second correlation using a pre-trained edge weight model. This edge weight model can be trained to have predictive capabilities using the second correlation as input and the edge weight as output. During training, sample second correlations and sample edge weights can be used as sample data, and a common training algorithm (e.g., gradient descent) can be employed to train the model, so that the trained model possesses the aforementioned capabilities.

[0059] In an optional implementation, constructing the power relationship graph based on the multiple nodes, the edges, and the weights of the edges includes: Constructing an initial relationship graph based on the multiple nodes, the edges, and the weights of the edges; Determining an initial functional node corresponding to each functional device in the initial relationship graph; Determining the functional criticality of each initial functional node on the initial relationship graph based on the weights of all edges connected to each initial functional node and the initial relationship graph; Based on the functional criticality of each initial functional node, the weights of all edges connected to each initial functional node are updated to update the initial relationship graph, and the updated initial relationship graph is used as the power relationship graph.

[0060] It can be understood that the higher the weight of the edge, the more important the association relationship between the initial function node connected by the edge and another node (such as functional synergy, spatial proximity, parameter consistency and / or historical behavior similarity) is. That is, no matter whether the function of the initial function node itself is important or not, its corresponding association relationship is also important. In this way, the functional criticality can be used to comprehensively measure the importance of the association relationship corresponding to each initial function node and the position in the graph. For high importance (i.e., high functional criticality), it means that the initial function node needs to be given extra attention, and thus the weights of all edges connected to the initial function node need to be increased. Correspondingly, for low importance (i.e., low functional criticality), you can choose to reduce the weights of all edges connected to the initial function node.

[0061] In some examples, determining the functional criticality of each initial functional node in the initial relationship graph based on the respective weights of all edges connected to each initial functional node and the initial relationship graph can include: obtaining a first functional criticality of each initial functional node by calculating a weighted average of the respective weights of all edges connected to each initial functional node; obtaining a second functional criticality of each initial functional node by determining a node degree corresponding to each initial functional node; and obtaining the functional criticality of each initial functional node in the initial relationship graph by performing a weighted calculation (e.g., weighted addition) on the first functional criticality and the second functional criticality. The node degree is determined by the number of nodes directly connected to the initial functional node.

[0062] In an optional implementation, the predicting the fault information corresponding to each switch node based on the first comprehensive feature information and the second vibration feature information includes: predicting a fault probability corresponding to each switch node based on the first comprehensive feature information and the second vibration feature information; For each switch node, if the fault probability corresponding to the switch node meets a preset condition, obtaining real-time monitoring data corresponding to the switch node, and determining the fault information corresponding to the switch node based on the real-time monitoring data and the second vibration feature information.

[0063] In some examples, the fault prediction model can be trained in advance to predict the fault probability (e.g., the probability of an abnormal vibration event) corresponding to each switch node based on the first comprehensive feature information and the second vibration feature information. The fault prediction model can be trained to have the ability to take the first comprehensive feature information and the second vibration feature information as input and take the fault probability as output. In the training process, sample comprehensive feature information and sample second vibration feature information can be used as sample data, which also carries an expected corresponding fault label representing the corresponding expected fault probability. A general training algorithm (e.g., gradient descent method) can be used to train the model, so that the trained model can have the above-mentioned ability.

[0064] In some examples, the fault prediction model can also be trained to have the ability to take real-time monitoring data and second vibration feature information as input and take corresponding fault information as output. The training process is not described here.

[0065] In some examples, the preset condition can include that the fault probability is higher than a preset fault probability threshold.

[0066] In this embodiment, the probability of each disconnector failure can be preliminarily determined by vibration data. When the probability is large, real-time monitoring data of the corresponding disconnector can be obtained for further fault analysis, thereby ensuring the accuracy of fault prediction while ensuring that computing power is not wasted.

[0067] In an optional embodiment, the real-time monitoring data corresponding to the switch node includes: line voltage fluctuation data and operating status data corresponding to the switch node, data of the isolating switch and adjacent devices corresponding to the switch node, and first environmental data of an area where the isolating switch corresponding to the switch node is located, wherein the adjacent device is a device adjacent to the isolating switch corresponding to the switch node among the multiple devices under test; The determining, based on the real-time monitoring data and the second vibration characterization information, fault information corresponding to the switch node includes: Determining a device state indicator characteristic of the switch node based on the line voltage fluctuation data and the second vibration characterization information; Determining a potential risk level of early failure based on the equipment status indicator characteristics; When the potential risk level of the early fault is higher than the preset level, the fault information corresponding to the switch node is determined based on the data of the disconnector and adjacent equipment corresponding to the switch node, as well as the first environment data and the operating status data.

[0068] In some examples, line voltage fluctuation data can be used to reflect whether there are electrical abnormalities in the operation of the equipment (such as poor contact, arc discharge, etc.); operating status data may include action time and / or current load; first environmental data may include temperature, humidity, dust concentration and / or electromagnetic interference intensity.

[0069] In some examples, the above-mentioned device status indicator characteristics can be obtained by performing feature extraction and feature fusion on the line voltage fluctuation data and the second vibration characterization information, so that the device status indicator characteristics can be used to indicate the line voltage fluctuation of the switch node, the vibration of the switch node, and the correlation coefficient between the vibration of the switch node and the voltage change.

[0070] In some examples, the device status indicator characteristics can be mapped to the corresponding early failure potential risk level through a mapping relationship table derived from experimental data, and the early failure potential risk level can also be determined based on the device status indicator characteristics through a pre-trained risk assessment model, wherein the risk assessment model can be a trained model with the ability to predict using the device status indicator characteristics as model input and the early failure potential risk level as model output. The specific training method can refer to the relevant model training method of this application and will not be repeated here.

[0071] In an optional embodiment, the data of the isolating switch and adjacent equipment corresponding to the switch node includes: accumulated data of thermal effects of the isolating switch and adjacent equipment corresponding to the switch node, and partial discharge data between the isolating switch and adjacent equipment corresponding to the switch node; The determining, based on the data of the isolating switch and adjacent equipment corresponding to the switch node, as well as the first environment data and the operating status data, the fault information corresponding to the switch node includes: Determining a fault range based on the thermal effect accumulation data and the partial discharge data; Determining second environmental data in the first environmental data that matches the fault range; Fault information corresponding to the switch node is determined based on the second environment data and the operating status data.

[0072] In some examples, the thermal effect accumulation data can be used to reflect the heat accumulation of the device during long-term operation.

[0073] In some examples, partial discharge (PD) is an indicator for assessing the insulation health of high-voltage electrical equipment. The PD data can be used to characterize the PD phenomenon between the disconnector corresponding to the switch node and the adjacent equipment. In addition, sensors installed on both devices can be used to detect PD signals in real time to generate the PD data.

[0074] In some examples, determining the fault scope based on the accumulated thermal effect data and the partial discharge data may include: comparing and analyzing the accumulated thermal effect data with historical accumulated thermal effect data to identify abnormally elevated temperature areas, where such abnormally elevated temperature areas typically indicate problems with the corresponding equipment, such as poor contact or overload; analyzing the partial discharge data to detect the presence of partial discharge signals, which, if present, indicate early signs of insulation degradation. Based on the location information of the partial discharge signals, the faulty device can be preliminarily located from the disconnector and adjacent devices corresponding to the switch node (for example, if a partial discharge signal is present between the disconnector and a certain adjacent device, it indicates that the disconnector and the adjacent device may be at fault and are thus the initially located faulty devices); and finally, determining the fault scope by combining the abnormally elevated temperature area and the located faulty devices. For example, if a certain adjacent device of the disconnector falls within both the abnormally elevated temperature area and the located faulty device, the area between the disconnector and the adjacent device falls within the determined fault scope. Consequently, second environmental data matching the fault scope can be subsequently used to accurately analyze the fault information corresponding to the switch node.

[0075] In some examples, determining the fault information corresponding to the switch node based on the second environment data and the operating status data may include: invoking a pre-trained fault prediction model to determine the fault information corresponding to the switch node based on the second environment data and the operating status data. The fault prediction model may be a trained model that has a predictive capability that takes the second environment data and the operating status data as input and outputs fault information.

[0076] The training method of the fault prediction model is as follows.

[0077] Sample fault data and its corresponding sample label are obtained, wherein the sample fault data includes second sample environment data and sample operation status data, and the sample label is used to indicate expected fault information.

[0078] The sample fault data is input into the model to be trained to obtain predicted fault information output by the model to be trained, wherein the model to be trained is constructed based on a deep learning algorithm and / or a machine learning algorithm.

[0079] Based on the difference between the predicted fault information and the expected fault information, a loss value is calculated by using the mean square error, and the model to be trained is trained using the loss value.

[0080] The above process is iterated until the calculated loss value is less than the preset loss value threshold, and a trained fault prediction model is obtained.

[0081] On the second aspect, accordingly, the embodiments of the present application also provide an isolating switch fault prediction system based on intelligent perception and data analysis, which can implement all the processes of the isolating switch fault prediction method based on intelligent perception and data analysis provided in the above embodiments.

[0082] See also Figure 2 , shows a structural diagram of an isolating switch fault prediction system based on intelligent perception and data analysis provided in an embodiment of the present application.

[0083] The present application provides an isolating switch fault prediction system based on intelligent perception and data analysis, including: A vibration data acquisition module 201 is configured to acquire vibration data of a plurality of devices under test through an intelligent sensor, wherein the plurality of devices under test include a plurality of isolating switches and functional devices between the isolating switches; A graph module 202 is configured to construct a power relationship graph corresponding to the plurality of devices under test, and determine in the power relationship graph the switch node corresponding to the isolation switch and the function node corresponding to the function device; A graph representation module 203, configured to determine first graph representation information of the functional node based on the graph information of the functional node on the power relationship graph; a vibration characterization module 204 for determining first vibration characterization information of the functional node and second vibration characterization information of the switch node based on the vibration data; A correlation module 205 is configured to determine a first correlation between the function node and the switch node in the power relationship graph; a comprehensive characterization module 206 for determining a function weight of the function node based on the first correlation and the device attribute of the function device, and combining the first graph characterization information and the first vibration characterization information based on the function weight to obtain first comprehensive characterization information of the function node; The prediction module 207 is configured to predict fault information corresponding to the switch node based on the first comprehensive representation information and the second vibration representation information.

[0084] In an optional implementation, determining the first correlation between the functional node and the switch node includes: For each functional node, Determine, among the switch nodes, a first switch node that is connected to the functional node; Determining a second switching node directly connected to the functional node from the first switching nodes; A first correlation corresponding to the functional node is determined based on the first switch node and the second switch node.

[0085] In an optional implementation, determining the first correlation corresponding to the functional node based on the first switch node and the second switch node includes: Determine all routes from the functional node to the first switch node; Based on the information of all the lines, the information of the first switch node, and the information of the second switch node, a first correlation corresponding to the functional node is determined.

[0086] In an optional implementation, constructing the power relationship graph corresponding to the plurality of devices under test includes: Obtaining device information of the multiple devices under test; Taking the multiple devices under test as multiple nodes, and determining each edge between the multiple nodes according to electrical connection information between the multiple devices under test; Determine the weight of each edge according to the device information; The power relationship graph is constructed based on the multiple nodes, the edges and the weights of the edges.

[0087] In an optional implementation, the determining the weight of each of the edges according to the device information comprises: for each of the edges, determining different nodes connected by the edge in the plurality of nodes, determining a second correlation degree of the different nodes based on the device information corresponding to each of the different nodes; determining the weight of the edge based on the second correlation degree.

[0088] In an optional implementation, the constructing the power relation graph based on the plurality of nodes, the edges and the weights of the edges comprises: constructing an initial relation graph based on the plurality of nodes, the edges and the weights of the edges; determining an initial function node corresponding to each function device in the initial relation graph; determining a function criticality degree of each initial function node in the initial relation graph based on the weight of each edge connected by the initial function node and the initial relation graph; updating the weight of each edge connected by each initial function node based on the function criticality degree of the initial function node, to update the initial relation graph, and taking the updated initial relation graph as the power relation graph.

[0089] In an optional implementation, the predicting the fault information corresponding to the switch node based on the first comprehensive representation information and the second vibration representation information comprises: predicting a fault probability corresponding to each switch node based on the first comprehensive representation information and the second vibration representation information; for each switch node, if the fault probability corresponding to the switch node satisfies a preset condition, obtaining real-time monitoring data corresponding to the switch node, and determining fault information corresponding to the switch node based on the real-time monitoring data and the second vibration representation information.

[0090] In an optional implementation, the real-time monitoring data corresponding to the switch node comprises line voltage fluctuation data and operating state data of the switch node, data of disconnectors and adjacent devices corresponding to the switch node, and first environmental data of an area where the disconnectors corresponding to the switch node are located, wherein the adjacent devices are devices adjacent to the disconnectors corresponding to the switch node in the plurality of devices to be tested. The determining the fault information corresponding to the switch node based on the real-time monitoring data and the second vibration representation information comprises: determine a device state index feature of the switching node based on the line voltage fluctuation data and the second vibration representation information; determine an early fault potential risk level based on the device state index feature; in a case where the early fault potential risk level is higher than a preset level, determine fault information corresponding to the switching node based on data of an isolating switch and adjacent devices corresponding to the switching node, and the first environment data and the operation state data.

[0091] In an optional implementation, the data of the isolating switch and the adjacent devices corresponding to the switching node include thermal effect accumulation data of the isolating switch and the adjacent devices corresponding to the switching node, and partial discharge data between the isolating switch and the adjacent devices corresponding to the switching node. The determining of the fault information corresponding to the switching node based on the data of the isolating switch and the adjacent devices corresponding to the switching node, and the first environment data and the operation state data includes: determining a fault range based on the thermal effect accumulation data and the partial discharge data; determining second environment data in the first environment data that matches the fault range; determining the fault information corresponding to the switching node based on the second environment data and the operation state data.

[0092] In a third aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement steps of the isolating switch fault prediction method based on intelligent sensing and data analysis.

[0093] In a fourth aspect, an embodiment of the present application provides a computer program product, which includes computer instructions. The computer instructions are executed by a processor to implement steps of the isolating switch fault prediction method based on intelligent sensing and data analysis.

[0094] In a fifth aspect, an embodiment of the present application provides a computer device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. The processor executes the computer program to implement steps of the isolating switch fault prediction method based on intelligent sensing and data analysis.

[0095] Reference is made to Figure 3The computer device of this embodiment includes: a processor 301, a memory 302, and a computer program stored in the memory 302 and executable on the processor 301, such as an isolating switch fault prediction program based on intelligent perception and data analysis. When the processor 301 executes the computer program, the steps of each of the above-mentioned isolating switch fault prediction method embodiments based on intelligent perception and data analysis are implemented, such as Figure 1 Steps S101-S107 are shown.

[0096] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 302 and executed by the processor 301 to implement the present application. The 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 in the computer device.

[0097] The computer device may be a desktop computer, laptop, PDA, cloud server, or other computing device. The computer device may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will appreciate that the schematic diagram is merely an example of a computer device and does not limit the computer device. The computer device may include more or fewer components than shown, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, and the like.

[0098] The processor 301 may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or the processor 301 may be any conventional processor. The processor 301 is the control center of the computer device, connecting various parts of the entire computer device using various interfaces and lines.

[0099] The memory 302 can be used to store the computer programs and / or modules. The processor 301 implements the various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 302 and accessing the data stored in the memory 302. The memory 302 may mainly include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as a sound playback function or an image playback function); the data storage area may store data generated based on the use of the mobile phone (such as audio data, a phone book, etc.). In addition, the memory 302 may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0100] If the module / unit integrated into the computer device is implemented as 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 application can implement all or part of the process in the above-mentioned embodiment method by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor 301, it can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the 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), electrical carrier signal, telecommunication signal, and software distribution medium.

[0101] In summary, the embodiments of the present application have at least the following beneficial effects: According to an embodiment of the present application, vibration data of multiple devices under test are obtained through an intelligent sensor, wherein the multiple devices under test include multiple isolating switches and functional devices between the isolating switches; a power relationship graph corresponding to the multiple devices under test is constructed, and the switch node corresponding to the isolating switch and the functional node corresponding to the functional device are determined in the power relationship graph; based on the graph information of the functional node on the power relationship graph, the first graph representation information of the functional node is determined; based on the vibration data, the first vibration representation information of the functional node and the second vibration representation information of the switch node are determined; in the power relationship graph, the first correlation between the functional node and the switch node is determined; based on the first correlation and the device attributes of the functional device, the functional weight of the functional node is determined, and the first graph representation information and the first vibration representation information are merged based on the functional weight to obtain the first comprehensive representation information of the functional node; based on the first comprehensive representation information and the second vibration representation information, the fault information corresponding to the switch node is predicted, so as to effectively deal with the risk of chain failures and improve the accuracy of isolating switch fault prediction.

[0102] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary hardware platform, and of course, it can also be implemented entirely through hardware. Based on this understanding, all or part of the contribution of the technical solution of the present application to the background technology can be embodied in the form of a software product. The computer software product can be stored in a storage medium, such as ROM (Read-Only Memory) / RAM (Random Access Memory), a disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application or certain parts of the embodiments.

[0103] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications are also considered to be within the scope of protection of the present application.

Claims

1. A disconnector fault prediction method based on intelligent perception and data analysis, characterized in that: include: Acquiring vibration data of a plurality of devices under test by means of an intelligent sensor, wherein the plurality of devices under test include a plurality of isolating switches and functional devices between the isolating switches; Constructing a power relationship graph corresponding to the plurality of devices under test, and determining a switch node corresponding to the isolation switch and a function node corresponding to the function device in the power relationship graph; Determining first graph representation information of the function node based on graph information of the function node on the power relationship graph; Determining first vibration characterization information of the functional node and second vibration characterization information of the switch node based on the vibration data; In the power relationship graph, determining a first correlation between the function node and the switch node; determining a function weight of the function node based on the first correlation and the device attribute of the function device, and combining the first graph representation information and the first vibration representation information based on the function weight to obtain first comprehensive representation information of the function node; Fault information corresponding to the switch node is predicted based on the first comprehensive representation information and the second vibration representation information.

2. The method according to claim 1, characterized in that The determining a first correlation between the functional node and the switch node includes: For each functional node, Determine, among the switch nodes, a first switch node that is connected to the functional node; Determining a second switching node directly connected to the functional node from the first switching nodes; A first correlation corresponding to the functional node is determined based on the first switch node and the second switch node.

3. The method according to claim 2, characterized in that The determining, based on the first switch node and the second switch node, a first correlation corresponding to the functional node includes: Determine all routes from the functional node to the first switch node; Based on the information of all the lines, the information of the first switch node, and the information of the second switch node, a first correlation corresponding to the functional node is determined.

4. The method according to claim 1, wherein The constructing of the power relationship diagram corresponding to the plurality of devices under test includes: Obtaining device information of the multiple devices under test; Taking the multiple devices under test as multiple nodes, and determining each edge between the multiple nodes according to electrical connection information between the multiple devices under test; Determine the weight of each edge according to the device information; The power relationship graph is constructed based on the multiple nodes, the edges and the weights of the edges.

5. The method according to claim 4, characterized in that Determining the weight of each edge according to the device information includes: For each of the edges, Determining different nodes connected by the edge among the multiple nodes, and determining a second correlation of the different nodes based on device information corresponding to each of the different nodes; Based on the second correlation, a weight of the edge is determined.

6. The method according to claim 4, characterized in that The constructing the power relationship graph based on the plurality of nodes, the edges, and the weights of the edges includes: Constructing an initial relationship graph based on the multiple nodes, the edges, and the weights of the edges; Determining an initial functional node corresponding to each functional device in the initial relationship graph; Determining the functional criticality of each initial functional node on the initial relationship graph based on the weights of all edges connected to each initial functional node and the initial relationship graph; Based on the functional criticality of each initial functional node, the weights of all edges connected to each initial functional node are updated to update the initial relationship graph, and the updated initial relationship graph is used as the power relationship graph.

7. The method according to claim 1, characterized in that The predicting, based on the first comprehensive representation information and the second vibration representation information, fault information corresponding to the switch node includes: Predicting a failure probability corresponding to each switch node based on the first comprehensive characterization information and the second vibration characterization information; For each switch node, if its corresponding fault probability meets a preset condition, real-time monitoring data corresponding to the switch node is obtained, and fault information corresponding to the switch node is determined based on the real-time monitoring data and the second vibration characterization information.

8. The method according to claim 7, characterized in that The real-time monitoring data corresponding to the switch node includes: line voltage fluctuation data and operating status data corresponding to the switch node, data of the isolating switch and adjacent devices corresponding to the switch node, and first environmental data of an area where the isolating switch corresponding to the switch node is located, wherein the adjacent devices are devices adjacent to the isolating switch corresponding to the switch node among the multiple devices under test; The determining, based on the real-time monitoring data and the second vibration characterization information, fault information corresponding to the switch node includes: Determining a device state indicator characteristic of the switch node based on the line voltage fluctuation data and the second vibration characterization information; Determining a potential risk level of early failure based on the equipment status indicator characteristics; When the potential risk level of the early fault is higher than the preset level, the fault information corresponding to the switch node is determined based on the data of the disconnector and adjacent equipment corresponding to the switch node, as well as the first environment data and the operating status data.

9. The method according to claim 8, characterized in that The data of the disconnector and adjacent equipment corresponding to the switch node include: thermal effect accumulation data of the disconnector and adjacent equipment corresponding to the switch node, and partial discharge data between the disconnector and adjacent equipment corresponding to the switch node; The determining, based on the data of the isolating switch and adjacent equipment corresponding to the switch node, as well as the first environment data and the operating status data, the fault information corresponding to the switch node includes: Determining a fault range based on the thermal effect accumulation data and the partial discharge data; Determining second environmental data in the first environmental data that matches the fault range; Fault information corresponding to the switch node is determined based on the second environment data and the operating status data.

10. A disconnector fault prediction system based on intelligent perception and data analysis, characterized in that: include: A vibration data acquisition module, configured to acquire vibration data of a plurality of devices under test through an intelligent sensor, wherein the plurality of devices under test include a plurality of isolating switches and functional devices between the isolating switches; A graph module, configured to construct a power relationship graph corresponding to the plurality of devices under test, and determine in the power relationship graph the switch node corresponding to the isolation switch and the function node corresponding to the function device; A graph representation module, configured to determine first graph representation information of the functional node based on graph information of the functional node on the power relationship graph; a vibration characterization module, configured to determine first vibration characterization information of the functional node and second vibration characterization information of the switch node based on the vibration data; A correlation module, configured to determine, in the power relationship graph, a first correlation between the function node and the switch node; a comprehensive characterization module, configured to determine a function weight of the function node based on the first correlation and the device attribute of the function device, and combine the first graph characterization information and the first vibration characterization information based on the function weight to obtain first comprehensive characterization information of the function node; A prediction module is used to predict fault information corresponding to the switch node based on the first comprehensive characterization information and the second vibration characterization information.

Images