Power distribution network power-related risk identification method, device, equipment, medium and product

Abstract

The invention provides a power distribution network electricity-related risk identification method and device, equipment, a medium and a product. The method comprises the following steps: acquiring electrical parameters, non-electrical parameters and environmental parameters of a plurality of pieces of power equipment in a power distribution network at a plurality of moments; for each moment, according to the non-electrical parameters of the multiple pieces of power equipment corresponding to the moment, adding non-electrical parameters in the first knowledge graph to obtain a second knowledge graph corresponding to the moment; obtaining an attribute feature corresponding to the moment and an association relationship between the power equipment and the attribute feature from a second knowledge graph corresponding to the moment; and inputting the electrical parameters, the environmental parameters and the attribute features at the plurality of moments and the association relationship between the power equipment and the attribute features into a pre-trained power-related risk identification model, and predicting to obtain the power-related risk condition of each power equipment in the plurality of power equipment. The method is used for achieving the effect of accurately predicting the potential electricity-related risk in the power distribution network.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of new power systems and artificial intelligence, and particularly relates to a power distribution network electrical risk identification method and device, equipment, medium and product. BACKGROUND

[0002] With the continuous optimization and upgrading of China's power structure and the continuous development of power innovation technology, the application of cutting-edge technologies such as cloud computing and big data in power grids is becoming more and more widespread. Using modern technology to do a good job in the operation and inspection of power distribution networks has become an important way to improve the reliability of power distribution networks. However, with the increasing proportion of distributed new energy (such as photovoltaic power generation, electric vehicle charging stations, and energy storage devices) connected to the power distribution network, the operation of the power distribution network is facing new challenges.

[0003] The connection of these distributed energy sources has changed the traditional power flow and operation mode, introducing a series of problems such as voltage fluctuations, frequency instability, and power reverse flow, further increasing the complexity and risk of power distribution network operation.

[0004] Therefore, in the face of such complex and variable operating environment, a new technology is needed to accurately predict potential electrical risks in the power distribution network. SUMMARY

[0005] The embodiments of the present application provide a power distribution network electrical risk identification method, device, equipment, medium and product, to accurately predict potential electrical risks in the power distribution network.

[0006] In a first aspect, the embodiments of the present application provide a power distribution network electrical risk identification method, comprising:

[0007] Collecting electrical parameters, non-electrical parameters and environmental parameters of a plurality of power devices in a power distribution network at a plurality of times;

[0008] For each time, according to the non-electrical parameters of the plurality of power devices corresponding to the time, adding the non-electrical parameters in the first knowledge graph to obtain the second knowledge graph corresponding to the time, the first knowledge graph comprising the connection relationship between the plurality of power devices, and the geographical position, device type and inherent attribute parameters of the plurality of power devices;

[0009] From the second knowledge graph corresponding to the time, extracting attribute features corresponding to the time and the association relationship between the power devices and the attribute features, the attribute features including static attribute features and dynamic attribute features; the dynamic attribute features are extracted from the non-electrical parameters in the second knowledge graph;

[0010] The electrical parameters, environmental parameters, attribute characteristics, and the association relationship between the power equipment and the attribute characteristics at multiple time points are input into a pre-trained electrical involvement risk identification model to predict the electrical involvement risk situation of each power equipment in the multiple power equipment; the electrical involvement risk situation is used to indicate whether the power equipment has an electrical involvement risk and the risk type when the electrical involvement risk exists.

[0011] In a possible implementation, the electrical involvement risk identification model comprises a knowledge fusion module, a graph convolution network, and a Transformer network, wherein:

[0012] The knowledge fusion module is configured to perform knowledge fusion on the electrical parameters, environmental parameters, and static attribute characteristics to obtain first fusion features, perform knowledge fusion on the electrical parameters, environmental parameters, and dynamic attribute characteristics to obtain second fusion features, and splice the first fusion features and the second fusion features to obtain third fusion features.

[0013] The graph convolution network is configured to perform spatial feature extraction on the third fusion features based on the association relationship between the power equipment and the attribute characteristics to obtain a new representation of the third fusion features.

[0014] The Transformer network is configured to capture a time sequence relationship in the new representation of the third fusion features corresponding to each time point in the multiple time points to obtain a current electrical involvement risk situation of each power equipment in the multiple power equipment.

[0015] In a possible implementation, the extracting the attribute characteristics corresponding to the time point and the association relationship between the power equipment and the attribute characteristics from the second knowledge graph corresponding to the time point comprises:

[0016] For each power equipment, a plurality of nodes connected to the power equipment and an association relationship between each node and the power equipment are obtained from the second knowledge graph corresponding to the time point; the content of the plurality of nodes comprises non-electrical parameters, other power equipment, geographical location, device type, and inherent attribute parameters associated with the power equipment.

[0017] The dynamic attribute characteristics corresponding to the time point are extracted according to each power equipment and the nodes connected to the power equipment and having non-electrical parameters as the attribute, and the static attribute characteristics corresponding to the time point are extracted according to each power equipment and the nodes connected to the power equipment and having other power equipment, geographical location, device type, and inherent attribute parameters as the attribute.

[0018] The association relationship between the power equipment and the attribute characteristics is obtained according to the association relationship between each node connected to each power equipment and the power equipment.

[0019] In a possible implementation, the electrical risk identification model is trained in the following manner:

[0020] Construct multiple sets of training sample data sets, each of which includes the electrical parameters, environmental parameters, attribute characteristics of multiple power devices at multiple times, the correlation between power devices and attribute characteristics, and the actual current electrical risk status of each power device;

[0021] Inputting electrical parameters, environmental parameters, attribute characteristics of multiple power devices at multiple times, and the correlation between the power devices and the attribute characteristics into the power-related risk identification model to be trained, and obtaining the current predicted power-related risk status of each of the multiple power devices;

[0022] Based on the cross entropy loss function, determine the difference between the predicted electricity-related risk situation and the actual electricity-related risk situation;

[0023] According to the difference, the model parameters of the electrical risk identification model to be trained are adjusted.

[0024] In a possible implementation, the multiple sets of training sample data sets are constructed in the following manner:

[0025] Acquire multiple sets of historical time series data, each of which includes the electrical parameters, non-electrical parameters, and environmental parameters of multiple power devices in the distribution network at multiple times, as well as the actual current electrical risk status of each power device;

[0026] For each moment in each set of historical time series data, based on the non-electrical parameters of multiple power devices at the moment in the historical time series data, the non-electrical parameters are added to the first knowledge graph to obtain a second knowledge graph corresponding to the moment;

[0027] Extracting, from the second knowledge graph corresponding to the moment, the attribute features corresponding to the moment and the association between the power equipment and the attribute features;

[0028] The multiple sets of training sample data sets are constructed based on the electrical parameters, environmental parameters, attribute characteristics of multiple power equipment in the distribution network at multiple times corresponding to each set of historical time series data, the correlation between the power equipment and the attribute characteristics, and the actual power-related risk conditions of each power equipment.

[0029] In a possible implementation, obtaining multiple sets of historical time series data includes:

[0030] Acquire multiple sets of time series data for multiple power devices in the distribution network, each set of time series data includes electrical parameters, non-electrical parameters, and environmental parameters;

[0031] For each set of time series data, the time series data is subjected to data cleaning and normalization processing to obtain processed time series data.

[0032] Based on the timestamps in the processed time series data, the electrical parameters, non-electrical parameters and environmental parameters in the processed time series data are subjected to timestamp alignment processing to obtain aligned time series data.

[0033] Expert knowledge is used to analyze the electrical parameters, non-electrical parameters and environmental parameters corresponding to each timestamp to determine the actual electricity-related risk situation corresponding to the timestamp.

[0034] Based on each set of aligned time series data and the actual electricity-related risk situation corresponding to each timestamp in the aligned time series data, the multiple sets of historical time series data are obtained.

[0035] In a second aspect, the embodiments of the present application provide a power distribution network electricity-related risk identification device, comprising:

[0036] The acquisition unit is configured to acquire electrical parameters, non-electrical parameters and environmental parameters of a plurality of power equipment in a power distribution network at a plurality of time points;

[0037] The setting unit is configured to, for each time point, add non-electrical parameters corresponding to the time point in a first knowledge graph based on the non-electrical parameters of the plurality of power equipment corresponding to the time point, to obtain a second knowledge graph corresponding to the time point, wherein the first knowledge graph comprises connection relationships between the plurality of power equipment, geographical positions, device types and inherent attribute parameters of the plurality of power equipment.

[0038] The acquisition unit is configured to extract attribute features corresponding to the time point and an association relationship between the power equipment and the attribute features from the second knowledge graph corresponding to the time point, wherein the attribute features include static attribute features and dynamic attribute features; and the dynamic attribute features are extracted from the non-electrical parameters in the second knowledge graph.

[0039] The prediction unit is configured to input the electrical parameters, environmental parameters, attribute features and association relationship between the power equipment and the attribute features at the plurality of time points into a pre-trained electricity-related risk identification model to predict an electricity-related risk situation of each power equipment in the plurality of power equipment; and the electricity-related risk situation is used to indicate whether the power equipment has an electricity-related risk and a risk type when the power equipment has an electricity-related risk.

[0040] In a possible implementation, the electricity-related risk identification model in the prediction unit comprises a knowledge fusion module, a graph convolution network and a Transformer network, wherein:

[0041] The knowledge fusion module is configured to fuse the electrical parameters, the environmental parameters, and the static attribute features to obtain first fused features, fuse the electrical parameters, the environmental parameters, and the dynamic attribute features to obtain second fused features, and splice the first fused features and the second fused features to obtain third fused features.

[0042] The graph convolution network is configured to perform spatial feature extraction on the third fused features based on the association between the power equipment and the attribute features to obtain a new representation of the third fused features.

[0043] The Transformer network is configured to capture a time sequence relationship in the new representation of the third fused features corresponding to each time point in the plurality of time points to obtain a current power-related risk situation of each power equipment in the plurality of power equipment.

[0044] In a possible implementation, the acquisition unit is configured to:

[0045] For each power equipment, the acquisition unit is configured to acquire, from the second knowledge graph corresponding to the time point, a plurality of nodes connected to the power equipment and an association between each node and the power equipment, and the content of the plurality of nodes includes non-electrical parameters, other power equipment, geographical location, device type, and inherent attribute parameters associated with the power equipment.

[0046] The acquisition unit is configured to extract, according to each power equipment and a node connected to the power equipment and having a non-electrical attribute, a dynamic attribute feature corresponding to the time point, and extract, according to each power equipment and a node connected to the power equipment and having an attribute of other power equipment, geographical location, device type, and inherent attribute parameter, a static attribute feature corresponding to the time point.

[0047] The acquisition unit is configured to acquire, according to the association between each node connected to each power equipment and the power equipment, an association between the power equipment and the attribute features.

[0048] In a possible implementation, the power grid power-related risk identification apparatus further includes:

[0049] The construction unit is configured to construct a plurality of sets of training sample data sets, and each set of training sample data set includes electrical parameters, environmental parameters, attribute features, an association between power equipment and attribute features, and an actual power-related risk situation of each power equipment of a plurality of power equipment at a plurality of time points.

[0050] The input unit is configured to input electrical parameters, environmental parameters, attribute characteristics, and an association relationship between the electrical parameters and the attribute characteristics of the plurality of electrical devices at the plurality of time points into the to-be-trained electrical risk identification model, to obtain a current predicted electrical risk situation of each electrical device in the plurality of electrical devices.

[0051] The determining unit is configured to determine a difference between the predicted electrical risk situation and an actual electrical risk situation based on a cross-entropy loss function.

[0052] The adjusting unit is configured to adjust model parameters of the to-be-trained electrical risk identification model according to the difference.

[0053] In a possible implementation, the constructing unit includes

[0054] The obtaining module is configured to obtain a plurality of groups of historical time series data, each group of historical time series data including electrical parameters, non-electrical parameters, environmental parameters, and an actual electrical risk situation of each electrical device in a plurality of electrical devices in a power distribution network at a plurality of time points.

[0055] The first processing module is configured to, for each time point in each group of historical time series data, add, in a first knowledge graph, a non-electrical parameter of each electrical device in the plurality of electrical devices at the time point according to the non-electrical parameter in the historical time series data, to obtain a second knowledge graph corresponding to the time point.

[0056] The second processing module is configured to extract, from the second knowledge graph corresponding to the time point, an attribute characteristic corresponding to the time point and an association relationship between the electrical device and the attribute characteristic.

[0057] The constructing module is configured to construct the plurality of groups of training sample data sets according to the electrical parameters, the environmental parameters, the attribute characteristics, the association relationship between the electrical device and the attribute characteristic, and the actual electrical risk situation of each electrical device of the plurality of electrical devices in the power distribution network at the plurality of time points corresponding to each group of historical time series data.

[0058] In a possible implementation, the obtaining module is specifically configured to:

[0059] The obtaining module is configured to obtain a plurality of groups of time series data of the plurality of electrical devices in the power distribution network, each group of time series data including electrical parameters, non-electrical parameters, and environmental parameters.

[0060] The obtaining module is configured to, for each group of time series data, perform data cleaning and normalization processing on the time series data to obtain processed time series data.

[0061] The obtaining module is configured to perform timestamp alignment processing on the electrical parameters, the non-electrical parameters, and the environmental parameters in the processed time series data based on timestamps in the processed time series data, to obtain aligned time series data.

[0062] The expert knowledge is used to analyze the electrical parameters, non-electrical parameters and environmental parameters corresponding to each timestamp, to determine the actual electrical risk situation corresponding to the timestamp.

[0063] Based on each set of aligned time series data and the actual electrical risk situation corresponding to each timestamp in the aligned time series data, the multiple sets of historical time series data are obtained.

[0064] In a third aspect, an embodiment of the present application provides a computer device, comprising: a memory, a processor;

[0065] The memory stores computer execution instructions.

[0066] The processor executes the computer execution instructions stored in the memory, so that the processor executes the first aspect and / or various possible implementation manners of the first aspect.

[0067] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed to implement the first aspect and / or various possible implementation manners of the first aspect.

[0068] In a fifth aspect, an embodiment of the present application provides a computer program product, comprising a computer program, which is executed to implement the first aspect and / or various possible implementation manners of the first aspect.

[0069] The power distribution network electrical risk identification method, device, equipment, medium and product provided by the embodiments of the present application enrich the feature information of the power equipment by predicting the electrical risk situation of each power equipment in the multiple power equipments based on the collected electrical parameters, non-electrical parameters and environmental parameters, and the connection relationship between the multiple power equipments, the geographical position, the equipment type and the inherent attribute parameters of the power equipments in the first knowledge graph, so that the pre-trained electrical risk identification model can more accurately predict, thereby achieving the effect of accurately predicting the potential electrical risk in the power distribution network. BRIEF DESCRIPTION OF DRAWINGS

[0070] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0071] Figure 1 A flowchart of the power distribution network electrical risk identification method provided for the first embodiment of the present application;

[0072] Figure 2 A structure diagram of the electrical risk identification model provided for the first embodiment of the present application;

[0073] Figure 3 A flowchart of a power distribution network electricity-related risk identification method provided for Embodiment Two of the present application is shown in FIG. 1.

[0074] Figure 4 A structural diagram of a power distribution network electricity-related risk identification device provided for Embodiment Three of the present application is shown in FIG. 2.

[0075] Figure 5 A structural diagram of a power distribution network electricity-related risk identification device provided for Embodiment Four of the present application is shown in FIG. 3.

[0076] Figure 6 A structural diagram of a computer device provided for the present application is shown in FIG. 4.

[0077] The above-described figures have shown specific embodiments of the present application, which will be described in more detail hereinafter. These figures and written descriptions are not intended to limit the scope of the concept of the present application in any way, but rather to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0078] The exemplary embodiments will be described in detail herein with reference to the attached drawings. The following description is with reference to the drawings, in which like numerals represent like elements, unless otherwise described in the following description. The embodiments described in the following exemplary embodiments are not representative of all embodiments consistent with the present application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the present application, as detailed in the appended claims.

[0079] In view of the technical problems in the background art, the inventors found, in the research on the power-related risk identification method of the power distribution network, that by first collecting electrical parameters (such as bus voltage and current of a bus transformer) of a plurality of power devices in the power distribution network at a plurality of time points, non-electrical parameters (such as temperature and on-off state of a photovoltaic inverter) and environmental parameters, then adding the non-electrical parameters to a first knowledge graph pre-acquired to include connection relationships between the plurality of power devices, geographical positions of the plurality of power devices, device types and inherent attribute parameters, a second knowledge graph is acquired. Based on this, the attribute features of the power distribution network at the plurality of time points and the association relationships between the power devices and the attribute features can be extracted from the second knowledge graph for extracting spatial features of the power distribution network at each time point. Further, by inputting the collected electrical parameters, environmental parameters, and the extracted attribute features and the association relationships between the power devices and the attribute features into a pre-trained power-related risk identification model, the current power-related risk situation of each power device in the plurality of power devices can be predicted. This method can enrich the feature information of the power devices, so that the pre-trained power-related risk identification model can more accurately predict, thereby achieving the effect of accurately predicting potential power-related risks in the power distribution network.

[0080] The technical solutions of the present application and how the technical solutions of the present application solve the above technical problems will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described again in some embodiments. The embodiments of the present application will be described below with reference to the accompanying drawings.

[0081] Figure 1 A flowchart of the power distribution network power-related risk identification method provided by Embodiment One of the present application is shown in FIG. 1, which includes the following steps. Figure 1

[0082] S101, collecting electrical parameters, non-electrical parameters and environmental parameters of a plurality of power devices in a power distribution network at a plurality of time points.

[0083] In this step, the electrical parameters, non-electrical parameters and environmental parameters of the plurality of power devices in the power distribution network are collected in real time to obtain the electrical parameters, non-electrical parameters and environmental parameters of the plurality of power devices at a plurality of time points.

[0084] The power device can be any power device connected to the power distribution network. For example, the power device can be a photovoltaic power generator, an electric vehicle charging station, an energy storage device, a bus transformer, a power distribution network line, a power distribution network node, etc. The selection of the power device is not specifically limited.

[0085] ​The electrical parameter of each power equipment refers to a parameter capable of being used to describe and characterize various electrical characteristics of the power equipment, which can dynamically reflect the operation condition and performance of the equipment. For example, the electrical parameter can include parameters such as voltage, current, output power, conversion efficiency, etc. of a photovoltaic inverter; parameters such as charging power, charging station load of an electric vehicle charging pile; parameters such as remaining power of a power storage device; parameters such as bus voltage, current, etc. of a bus transformer; parameters such as voltage, current, power, power factor, etc. of a distribution network line; and parameters such as load power of a distribution network node. The selection of the electrical parameter is not limited in the present application, and is limited to the above-mentioned electrical parameters.

[0086] The non-electrical parameter of each power equipment refers to a parameter that is not directly related to the electrical characteristics of the power equipment, but has an important influence on the operation state, performance evaluation, safety monitoring, etc. of the power equipment. For example, the non-electrical parameter can be parameters such as operation time, start state, temperature, etc. of the power equipment. The selection of the non-electrical parameter is not limited in the present application, and is limited to the above-mentioned non-electrical parameters.

[0087] The environmental parameter of each power equipment includes various parameters of the environment in which the power equipment is located, which will affect the operation stability of the power equipment. For example, the environmental parameter can include parameters such as temperature, humidity, wind speed, solar radiation, pollutant concentration, rainfall, etc. of the environment in which the power equipment is located. The selection of the environmental parameter is not limited in the present application, and is limited to the above-mentioned environmental parameters.

[0088] In S102, for each time point, the non-electrical parameters of the plurality of power equipment corresponding to the time point are added to the first knowledge graph to obtain a second knowledge graph corresponding to the time point, wherein the first knowledge graph includes the connection relationship between the plurality of power equipment, and the geographical position, equipment type and inherent attribute parameter of the plurality of power equipment.

[0089] In this step, for each time point, the non-electrical parameters of the plurality of power equipment corresponding to the time point are added to the first knowledge graph including the static characteristics of the distribution network to obtain a second knowledge graph corresponding to the time point. Based on the second knowledge graph, the electrical risk of each power equipment in the distribution network can be judged from the spatial point of view.

[0090] The static characteristics refer to the connection relationship between the plurality of power equipment, and the geographical position, equipment type and inherent attribute parameter of the plurality of power equipment. The inherent attribute parameter refers to a characteristic parameter determined by the design, manufacture and material of the power equipment, which does not change with the external operating conditions, for example, the rated capacity of the energy storage device and the rated transformation ratio of the bus transformer.

[0091] It should be understood that the connection relationship between the plurality of power equipment in the power distribution network, and the geographical position of the plurality of power equipment, provide a basis for judging the electrical risk of each power equipment in the power distribution network from a spatial perspective. In addition, the device type and inherent attribute parameters help to more accurately judge the electrical risk of each power equipment. Since the connection relationship between the plurality of power equipment, and the geographical position of the plurality of power equipment, the device type and the inherent attribute parameters are time-invariant data, a first knowledge graph can be directly constructed in advance, and in the subsequent prediction process, only the dynamically changing non-electrical parameters need to be added to the first knowledge graph, and a second knowledge graph obtained can be used to judge the electrical risk of each power equipment in the power distribution network from a spatial perspective.

[0092] It is worth noting that the most basic unit of constructing a knowledge graph is a triple, which is used to represent the relationship between entities in the knowledge graph or the relationship between attributes and attribute values, wherein the entities and attributes in the knowledge graph can be called nodes.

[0093] For example, the triples in the first knowledge graph include: "energy storage device A11, connected to, bus transformer B2" and "photovoltaic inverter F1, connected to, distribution line S3" and the like, which are the relationship between entities, wherein "energy storage device A11", "bus transformer B2", "photovoltaic inverter F1" and "distribution line S3" represent entities, and "connected to" represents the association relationship between entities.

[0094] The triples in the first knowledge graph also include: "electric vehicle charging pile U1, location, A area", "photovoltaic inverter F1, device type, photovoltaic device" and "energy storage device A11, rated capacity, 500kWh" and the like, which are the relationship between entities and attributes, wherein "electric vehicle charging pile U1", "photovoltaic inverter F1" and "energy storage device A11" represent entities, "A area", "photovoltaic device" and "500kWh" represent attributes, and "location", "device type" and "rated capacity" represent the association relationship between entities and attributes.

[0095] For another example, the triples added in the second knowledge graph can include: "energy storage device A11, start state, not started" and the like.

[0096] In S103, attribute features corresponding to the moment and the association relationship between the power equipment and the attribute features are extracted from the second knowledge graph corresponding to the moment, wherein the attribute features include static attribute features and dynamic attribute features.

[0097] Among them, the dynamic attribute features are extracted from the non-electrical parameters in the second knowledge graph.

[0098] In this step, for each time, the attribute features corresponding to the time and the association relationship between the power equipment and the attribute features need to be extracted from the second knowledge graph corresponding to the time. The attribute features and the association relationship between the power equipment and the attribute features are in the form of a matrix.

[0099] In actual application, the attribute features extracted from the second knowledge graph are a feature vector matrix, and the association relationship between the power equipment and the attribute features extracted from the second knowledge graph is an adjacency matrix. The feature vector matrix and the adjacency matrix will be used as the input of the power-related risk identification model. Correspondingly, the feature vector matrix includes two: a static feature matrix and a dynamic feature matrix.

[0100] In a possible implementation, the attribute features corresponding to the time and the association relationship between the power equipment and the attribute features can be extracted in the following manner:

[0101] Step 1, for each power equipment, the multiple nodes connected to the power equipment and the association relationship between each node and the power equipment are obtained from the second knowledge graph corresponding to the time. The content of the multiple nodes includes non-electrical parameters, other power equipment, geographical location, device type and inherent attribute parameters associated with the power equipment.

[0102] In this step, the nodes connected to each power equipment need to be obtained from the knowledge graph, wherein the nodes include entities and attributes in the second knowledge graph. In addition, the association relationship between each power equipment and the nodes connected thereto needs to be obtained, including the association relationship between entities and entities in the second knowledge graph and the association relationship between entities and attributes.

[0103] Step 2, according to each power equipment and the nodes connected to the power equipment and having non-electrical parameters as attributes, the dynamic attribute features corresponding to the time are extracted, and according to each power equipment and the nodes connected to the power equipment and having other power equipment, geographical location, device type and inherent attribute parameters as attributes, the static attribute features corresponding to the time are extracted.

[0104] In the specific implementation of the scheme, the dynamic attribute features corresponding to the time need to be extracted according to each power equipment and the nodes connected to the power equipment and having non-electrical parameters as attributes, to obtain a dynamic feature matrix. The rows and columns of the dynamic feature matrix are respectively the multiple power equipment and the nodes connected to each power equipment and having non-electrical parameters as attributes, and the matrix elements are the vector representation of the content of each node corresponding to each power equipment.

[0105] Similarly, the static attribute features corresponding to the moment are also needed to be extracted according to each power device and the nodes connected by the power device, which are attributed to other power devices, geographical locations, device types and inherent attribute parameters, to obtain a static feature matrix. The rows and columns of the static feature matrix are respectively the plurality of power devices and the nodes connected by each power device, which are attributed to other power devices, geographical locations, device types and inherent attribute parameters, and the matrix elements are the vector representation of the content of each node corresponding to each power device.

[0106] Step 3. Extract the association relationship between the power device and the attribute feature according to the association relationship between each node connected by each power device and the power device.

[0107] In the specific implementation of the present scheme, the association relationship between the power device and the attribute feature is an adjacency matrix. Wherein, the rows and columns of the adjacency matrix are respectively the plurality of power devices and the nodes connected by each power device, and the matrix elements represent the connection relationship between the two.

[0108] S104. Input the electrical parameters, environmental parameters, attribute features and association relationship between the power device and the attribute feature at multiple moments into the pre-trained electrical risk identification model to predict the current electrical risk situation of each power device in the plurality of power devices.

[0109] In this step, the electrical parameters, environmental parameters, attribute features and association relationship between the power device and the attribute feature at multiple moments obtained need to be input into the pre-trained electrical risk identification model to predict the current electrical risk situation of each power device in the plurality of power devices. Wherein, the electrical risk situation is used to indicate whether the power device has an electrical risk, and the risk type when there is an electrical risk.

[0110] For example, the electrical risk situation of the power device can be "no risk", "device overheating risk", "device circuit breaking risk" and "device overvoltage risk", etc.

[0111] In a specific implementation manner, Figure 2 The structural diagram of the electrical risk identification model provided by Embodiment One of the present application is shown in Figure 2 As shown, the electrical risk identification model includes a knowledge fusion module, a graph convolutional network (GCN) and a Transformer network.

[0112] The knowledge fusion module is configured to fuse the electrical parameters, the environmental parameters and the static attribute features to obtain a first fusion feature, fuse the electrical parameters, the environmental parameters and dynamic attribute features to obtain a second fusion feature, and splice the first fusion feature and the second fusion feature to obtain a third fusion feature.

[0113] Specifically, the knowledge fusion module is capable of fusing the electrical parameters, the environmental parameters and the static attribute features of the plurality of power equipment in the power distribution network at each time point.

[0114] In a specific implementation of the scheme, the electrical parameters and the environmental parameters are combined into a power grid feature X t together with static feature attribute features (i.e., a static feature matrix) e s and dynamic feature attribute features (i.e., a dynamic feature matrix) e d as inputs of the knowledge fusion module.

[0115] After receiving the above data, the knowledge fusion module performs the following operations:

[0116] ① fuse the power grid feature X t with the static feature attribute features e s to obtain a first fusion feature X s , and the calculation formula is as follows:

[0117] X s =ReLU(e s X t w s +b s )

[0118] In the above formula, w s represents a static weight matrix; b s represents a static bias term; and ReLU is a nonlinear activation function.

[0119] fuse the power grid feature X t with the dynamic feature attribute features e d to obtain a first fusion feature X d , and the calculation formula is as follows:

[0120] X d =ReLU(e d X t w d +b d )

[0121] In the above formula, w d represents a dynamic weight matrix; and b d represents a dynamic bias term.

[0122] ③ the first fusion feature X s and the second fusion feature X d are spliced to obtain a third fusion feature X The following formula is used to represent:

[0123]

[0124] It should be understood that the third fusion feature obtained through the above knowledge fusion operation of the knowledge fusion module includes not only the actual observation data such as the electrical parameters and the environmental parameters, but also the attribute features. The knowledge fusion module is used to realize seamless fusion of the knowledge graph and the subsequent neural network (graph convolution network and Transformer network), effectively integrating the static attribute features and the dynamic attribute features with the actual feature data of the power grid, and outputting more accurate and meaningful feature representations. This fusion mechanism not only enriches the input information of the subsequent neural network, but also improves the explainability of the electricity-related risk identification model.

[0125] In addition, the graph convolution network is used to perform spatial feature extraction on the third fusion feature based on the association relationship between the power equipment and the attribute features, to obtain spatial features of the plurality of power equipment at the corresponding time.

[0126] In the specific implementation of the present solution, the electricity-related risk identification model can include an l-layer CGN structure. The first-layer GCN receives the adjacency matrix (i.e., the association relationship between the power equipment and the attribute features) corresponding to any time, adds self-connection and the third fusion feature output by the knowledge fusion module, then aggregates the third fusion feature through the adjacency matrix, and linearly transforms and non-linearly activates the aggregated feature vector to obtain a new feature matrix. Correspondingly, in each subsequent GCN, the received new feature matrix is continuously processed according to the adjacency matrix until the last layer outputs the final new representation of the third fusion feature. The adjacency matrix plus self-connection means that a matrix obtained by adding the relationship between each node and itself to the original adjacency matrix.

[0127] The calculation formula for the lth-layer GCN to process the received new feature matrix according to the adjacency matrix is as follows:

[0128]

[0129] In the above formula, H (l+1) represents the new feature matrix output by the lth-layer GCN; represents the adjacency matrix plus self-connection; represents the degree matrix of , and when i≠j, W (l) represents the weight matrix of the lth-layer GCN; H (l)Xl represents a feature matrix of the GCN input of the lth layer; and σ represents a nonlinear activation function (for example, a ReLU function).

[0130] It should be understood that, based on the association relationship between the power equipment and the attribute features extracted from the knowledge graph, and the third fusion features fused with the multi-aspect data, the GCN can capture the dependency relationship between the power equipment in the power distribution network, and extract the spatial features between the power equipment, and finally input the new representation containing the third fusion features fused with the spatial features.

[0131] In addition, the Transformer network is used to capture the new representation of the third fusion features in each time point in the multiple time points, to obtain the current power-related risk situation of each power equipment in the multiple power equipment.

[0132] In the actual application of the present scheme, the Transformer network is a deep learning architecture based on an attention mechanism. Based on the new representation of the third fusion features in each time point output by the graph convolutional network, the Transformer network will capture the long-term dependency relationship in this time series data, to obtain new features fused with time features, and determine whether each power equipment has a power-related risk and the probability of the risk type when there is a power-related risk, so as to determine the power-related risk situation of each power equipment.

[0133] The power distribution network power-related risk identification method provided in the embodiment predicts the power-related risk situation of each power equipment in the multiple power equipment based on the collected electrical parameters, non-electrical parameters, and environmental parameters, and the connection relationship between the multiple power equipment pre-set in the first knowledge graph, the geographical position, the equipment type, and the inherent attribute parameters of the multiple power equipment, enriches the feature information of the power equipment, enables the pre-trained power-related risk identification model to more accurately predict, and thus achieves the effect of accurately predicting the potential power-related risk in the power distribution network. In addition, based on the power-related risk identification model provided in the embodiment, the electrical parameters, environmental parameters, and attribute features are first fused by the knowledge fusion module, then the spatial features are extracted by the graph convolutional network, and then the time features are extracted by the Transformer network, so as to more comprehensively and deeply understand the potential risk of the power distribution network, and further improve the accuracy of predicting the potential power-related risk in the power distribution network.

[0134] Figure 3 The flowchart of the power distribution network power-related risk identification method provided in Embodiment Two of the present application is shown in Figure 3 As shown in the above embodiment, the present embodiment further includes a training method of a power-related risk identification model, comprising:

[0135] S201, construct a plurality of training sample data sets, wherein each training sample data set includes electrical parameters, environmental parameters, attribute characteristics, an association between the electrical parameters and the attribute characteristics, and an actual electrical risk of each electrical device at a plurality of time points.

[0136] In this step, a plurality of historical data sets are needed to construct the training sample data set, each historical data set including electrical parameters, environmental parameters, and non-electrical parameters of a plurality of electrical devices at a plurality of time points.

[0137] In one possible implementation, the plurality of training sample data sets are constructed by the following steps 1-4:

[0138] Step 1, obtain a plurality of historical time series data, each historical time series data including electrical parameters, non-electrical parameters, environmental parameters, and an actual electrical risk of each electrical device in a plurality of time points in a power distribution network.

[0139] In the specific implementation of the present scheme, step 1 can specifically include the following steps 1.1-1.5:

[0140] Step 1.1, obtain a plurality of time series data of a plurality of electrical devices in a power distribution network, each time series data including electrical parameters, non-electrical parameters, and environmental parameters.

[0141] In this step, a plurality of electrical parameter time series data, non-electrical parameter time series data, and environmental parameter time series data are collected by sensors and other devices. Each time series data includes the numerical value of the electrical parameter / non-electrical parameter / environmental parameter and the corresponding timestamp.

[0142] Step 1.2, for each time series data, data cleaning and normalization processing are performed on the time series data to obtain processed time series data.

[0143] In this step, the time series data obtained in step 1.1 is subjected to data cleaning and normalization processing to obtain processed time series data. Data cleaning includes missing data processing, outlier processing, and redundant data processing to ensure data integrity and consistency; normalization processing is used to standardize data from different sources and dimensions to facilitate model analysis.

[0144] Step 1.3, based on the timestamps in the processed time series data, the electrical parameters, non-electrical parameters, and environmental parameters in the processed time series data are subjected to timestamp alignment processing to obtain aligned time series data.

[0145] In this step, the processed time series data is subjected to timestamp alignment processing to ensure the consistency of multi-source data.

[0146] Step 1.4, analyze the electrical parameters, non-electrical parameters and environmental parameters corresponding to each timestamp by using expert knowledge to determine the actual electrical risk situation corresponding to the timestamp.

[0147] In this step, it is necessary to analyze the electrical parameters, non-electrical parameters and environmental parameters corresponding to each timestamp by using expert knowledge to determine the actual electrical risk situation corresponding to the timestamp. Among them, the expert knowledge is the risk judgment rule determined by the electrical experts based on a large amount of practical experience and knowledge.

[0148] For example, the expert knowledge is, for example, after thunderstorm weather, if there is a lightning record in the area where the power distribution network is located, and part of the line appears instantaneous power failure or voltage fluctuation phenomenon, the expert will judge that this part of the line may be struck by lightning, and there is a risk of insulation damage, lightning arrester failure, etc.

[0149] It should be understood that each timestamp in the aligned time series data corresponds to each time in the historical time series data.

[0150] Step 1.5, based on each set of aligned time series data and the actual electrical risk situation corresponding to each timestamp in the aligned time series data, obtain a plurality of sets of historical time series data.

[0151] In this step, each set of aligned time series data includes electrical parameters, non-electrical parameters and environmental parameters of a plurality of power equipment in the power distribution network at a plurality of time points; the actual electrical risk situation corresponding to each timestamp in the aligned time series data also corresponds to the actual electrical risk situation of a plurality of power equipment in the power distribution network at a plurality of time points. The electrical parameters, non-electrical parameters and environmental parameters of the plurality of power equipment in the power distribution network at a plurality of time points are the plurality of sets of historical time series data.

[0152] The present implementation utilizes expert knowledge to analyze the electrical parameters, non-electrical parameters and environmental parameters corresponding to each timestamp to determine the actual electrical risk situation corresponding to the timestamp, effectively ensuring the reliability of the sample data set, and further improving the accuracy of the electrical risk identification model obtained by training.

[0153] Step 2, for each time in each set of historical time series data, according to the non-electrical parameters of the plurality of power equipment at the time in the historical time series data, increase the non-electrical parameters in the first knowledge graph to obtain the second knowledge graph corresponding to the time.

[0154] In this step, the non-electric parameters of the plurality of power equipment corresponding to each time point are added to the first knowledge graph to obtain a second knowledge graph corresponding to the time point, so as to extract the attribute characteristics corresponding to the time point and the association relationship between the power equipment and the attribute characteristics. The first knowledge graph includes the connection relationship between the plurality of power equipment, and the geographical position, equipment type and inherent attribute parameters of the plurality of power equipment.

[0155] Step 3, from the second knowledge graph corresponding to the time point, the attribute characteristics corresponding to the time point and the association relationship between the power equipment and the attribute characteristics are extracted.

[0156] Step 4, according to the electrical parameters, environmental parameters, attribute characteristics, association relationship between the power equipment and the attribute characteristics of the plurality of power equipment in the power distribution network corresponding to each group of historical time series data at a plurality of time points, and the actual power-related risk situation of each power equipment, a plurality of training sample data sets are constructed.

[0157] In this step, a plurality of training sample data sets can be constructed according to the electrical parameters, environmental parameters, attribute characteristics, association relationship between the power equipment and the attribute characteristics of the plurality of power equipment in the power distribution network at a plurality of time points, and the actual power-related risk situation of each power equipment.

[0158] S202, input the electrical parameters, environmental parameters, attribute characteristics, association relationship between the power equipment and the attribute characteristics of the plurality of power equipment at a plurality of time points into the power-related risk identification model to be trained, to obtain the current predicted power-related risk situation of each power equipment in the plurality of power equipment.

[0159] In this step, when training the power-related risk identification model to be trained using any sample data set, the electrical parameters, environmental parameters, attribute characteristics, association relationship between the power equipment and the attribute characteristics of the plurality of time points in the sample data set are input into the power-related risk identification model to be trained, to obtain the current predicted power-related risk situation of each power equipment in the plurality of power equipment.

[0160] S203, based on the cross-entropy loss function, the difference between the predicted power-related risk situation and the actual power-related risk situation is determined.

[0161] In this step, the cross-entropy loss function is used to calculate the difference between the predicted power-related risk situation and the actual power-related risk situation.

[0162] It should be understood that the Transformer network ultimately performs a classification task to predict the probability of the input sample data belonging to each category. For example, the categories can include "no risk for each power equipment" and "power equipment 12 has a short circuit risk, power equipment 3 has an insulation damage risk, and the rest of the power equipment has no risk".

[0163] Assuming there are C categories in total, the calculation formula of the difference of the cross-entropy loss function is as follows:

[0164]

[0165] In the above formula, denotes the probability that the sample predicted by the electrical risk model belongs to category c; y c denotes the indicator function of the true label on category c, when the true category of the sample is c, y c is 1, otherwise 0, where the true label is the current actual electrical risk of each power equipment.

[0166] S204, according to the difference, adjusting the model parameters of the electrical risk identification model to be trained.

[0167] In this step, it is necessary to adjust the model parameters of the electrical risk identification model to be trained based on the above difference, and with the input of the sample data set, the parameters in the model are continuously optimized.

[0168] In the electrical risk identification method provided by the embodiment, a plurality of training sample data sets including electrical parameters, environmental parameters, attribute characteristics, the association relationship between the power equipment and the attribute characteristics, and the current actual electrical risk of each power equipment are constructed in advance, and the electrical risk identification model to be trained is trained based on the cross-entropy loss function. The electrical risk identification model is successfully obtained, which provides a basis for the application of the model.

[0169] Figure 4 The structure diagram of the power distribution network electrical risk identification device provided by Embodiment Three of the present application is shown in Figure 4 As shown in the figure, the power distribution network electrical risk identification device 30 provided by the embodiment includes:

[0170] The acquisition unit 301 is configured to acquire electrical parameters, non-electrical parameters and environmental parameters of a plurality of power equipment in a power distribution network at a plurality of time points, respectively.

[0171] The setting unit 302 is configured to, for each time point, add non-electrical parameters of a plurality of power equipment corresponding to the time point in the first knowledge graph to obtain a second knowledge graph corresponding to the time point, wherein the first knowledge graph includes the connection relationship between the plurality of power equipment, and the geographical position, equipment type and inherent attribute parameters of the plurality of power equipment.

[0172] The acquisition unit 303 is configured to extract, from the second knowledge graph corresponding to the time point, attribute features corresponding to the time point and an association relationship between the power equipment and the attribute features, the attribute features including static attribute features and dynamic attribute features, and the dynamic attribute features being obtained by extracting non-electric parameters in the second knowledge graph;

[0173] The prediction unit 304 is configured to input the electric parameters, the environmental parameters, the attribute features, and the association relationship between the power equipment and the attribute features at the plurality of time points into a pre-trained electrical risk identification model to predict an electrical risk situation of each of the plurality of power equipment, and the electrical risk situation is used to indicate whether the power equipment has an electrical risk and a risk type when the power equipment has the electrical risk.

[0174] The power distribution network electrical risk identification device 30 provided in this embodiment can execute the method provided in the method embodiments, and has similar implementation principles and technical effects, which will not be described here in detail.

[0175] Figure 5 FIG. 3 shows a structural schematic diagram of the power distribution network electrical risk identification device provided in Embodiment Four of the present application. Figure 5 As shown in the above embodiment, the power distribution network electrical risk identification device 30 provided in this embodiment further includes:

[0176] The construction unit 305 is configured to construct a plurality of sets of training sample data sets, each set of training sample data set including the electric parameters, the environmental parameters, the attribute features, the association relationship between the power equipment and the attribute features, and the actual electrical risk situation of each power equipment at the plurality of time points.

[0177] The input unit 306 is configured to input the electric parameters, the environmental parameters, the attribute features, and the association relationship between the power equipment and the attribute features of the plurality of power equipment at the plurality of time points into the electrical risk identification model to be trained to obtain a current predicted electrical risk situation of each of the plurality of power equipment.

[0178] The determination unit 307 is configured to determine a difference between the predicted electrical risk situation and the actual electrical risk situation based on a cross-entropy loss function.

[0179] The adjustment unit 308 is configured to adjust model parameters of the electrical risk identification model to be trained according to the difference.

[0180] In a possible implementation, the electrical risk identification model in the prediction unit 304 includes a knowledge fusion module, a graph convolution network, and a Transformer network, wherein:

[0181] The knowledge fusion module is configured to fuse the electrical parameters, the environmental parameters, and the static attribute features to obtain first fused features; fuse the electrical parameters, the environmental parameters, and the dynamic attribute features to obtain second fused features; and splice the first fused features and the second fused features to obtain third fused features;

[0182] The graph convolution network is configured to perform spatial feature extraction on the third fused features based on the association relationship between the power equipment and the attribute features to obtain a new representation of the third fused features;

[0183] The Transformer network is configured to capture a time sequence relationship in the new representation of the third fused features corresponding to each time point in the plurality of time points to obtain a current electricity-related risk situation of each power equipment in the plurality of power equipment.

[0184] In a possible implementation, the acquisition unit 303 is configured to:

[0185] For each power equipment, the acquisition unit 303 is configured to acquire, from the second knowledge graph corresponding to the time point, a plurality of nodes connected to the power equipment and an association relationship between each node and the power equipment; and the content of the plurality of nodes includes non-electrical parameters, other power equipment, geographical positions, device types, and inherent attribute parameters associated with the power equipment.

[0186] The acquisition unit 303 is configured to extract, according to each power equipment and a node connected to the power equipment and having a non-electrical attribute, a dynamic attribute feature corresponding to the time point, and extract, according to each power equipment and a node connected to the power equipment and having an attribute of other power equipment, a geographical position, a device type, or an inherent attribute parameter, a static attribute feature corresponding to the time point.

[0187] The acquisition unit 303 is configured to acquire, according to the association relationship between each node connected to each power equipment and the power equipment, an association relationship between the power equipment and the attribute features.

[0188] In a possible implementation, the construction unit 305 includes

[0189] The acquisition module is configured to acquire a plurality of groups of historical time sequence data, each group of historical time sequence data including electrical parameters, non-electrical parameters, environmental parameters, and actual electricity-related risk situations of a plurality of power equipment in a power distribution network at a plurality of time points.

[0190] The first processing module is configured to, for each time point in each group of historical time sequence data, add, according to non-electrical parameters of the plurality of power equipment at the time point in the historical time sequence data, the non-electrical parameters in a first knowledge graph to obtain a second knowledge graph corresponding to the time point.

[0191] The second processing module is configured to extract the attribute features corresponding to the time point and the association between the power equipment and the attribute features from a second knowledge graph corresponding to the time point.

[0192] The construction module is configured to construct the multiple sets of training sample data sets according to the electrical parameters, the environmental parameters, the attribute features, the association between the power equipment and the attribute features, and the actual power-related risk situations of the multiple power equipment in the power distribution network at the multiple time points corresponding to each set of historical time series data.

[0193] In a possible implementation, the acquisition module is specifically configured to:

[0194] The acquisition module is configured to acquire multiple sets of time series data of multiple power equipment in a power distribution network, each set of time series data including electrical parameters, non-electrical parameters, and environmental parameters.

[0195] The acquisition module is configured to perform data cleaning and normalization processing on each set of time series data to obtain processed time series data.

[0196] The acquisition module is configured to perform timestamp alignment processing on the electrical parameters, the non-electrical parameters, and the environmental parameters in the processed time series data based on the timestamps in the processed time series data to obtain aligned time series data.

[0197] The acquisition module is configured to analyze the electrical parameters, the non-electrical parameters, and the environmental parameters corresponding to each timestamp using expert knowledge to determine the actual power-related risk situation corresponding to the timestamp.

[0198] The acquisition module is configured to obtain the multiple sets of historical time series data based on each set of aligned time series data and the actual power-related risk situation corresponding to each timestamp in the aligned time series data.

[0199] The power distribution network power-related risk identification device 30 provided in this embodiment can execute the method provided in the method embodiments, and has similar implementation principles and technical effects, which will not be described here in detail.

[0200] Figure 6 The computer device provided in this application is shown in a structural schematic diagram. Figure 6 As shown in the structural schematic diagram, the computer device 40 provided in this embodiment includes at least one processor 401 and a memory 402. Optionally, the computer device 40 further includes a communication component 403. The processor 401, the memory 402, and the communication component 403 are connected through a bus 404.

[0201] In the specific implementation process, the at least one processor 401 executes the computer execution instructions stored in the memory 402, so that the at least one processor 401 executes the method described above.

[0202] The specific implementation process of the processor 401 can refer to the method embodiments described above, which have similar implementation principles and technical effects. Details are not described here again.

[0203] In the above embodiments, it should be understood that the processor can be a central processing unit (English: Central Processing Unit, CPU for short), and can also be other general-purpose processors, digital signal processors (English: Digital Signal Processor, DSP for short), application specific integrated circuits (English: Application Specific Integrated Circuit, ASIC for short), etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in combination with the application can be directly embodied as hardware processor execution, or executed by hardware and software modules in the processor.

[0204] The memory can include read-only memory and random access memory. The memory can be volatile memory or non-volatile memory, or can include both volatile and non-volatile memory. Among them, the non-volatile memory can include read-only memory (Read-Only Memory, ROM), programmable read-only memory (Programmable ROM, PROM), erasable programmable read-only memory (Erasable PROM, EPROM), electrically erasable programmable read-only memory (Electrically EPROM, EEPROM) or flash memory. The volatile memory can include random access memory (Random Access Memory, RAM) used as an external cache. By way of example but not limitation, many forms of RAM can be used. For example, static random access memory (Static RAM, SRAM), dynamic random access memory (Dynamic Random Access Memory, DRAM), synchronous dynamic random access memory (Synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (Double Data Rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (Enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (Sync link DRAM, SLDRAM) and direct memory bus random access memory (Direct Rambus RAM, DR RAM).

[0205] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.

[0206] The present application also provides a computer program product comprising a computer program which, when executed, implements the power distribution network electrical risk identification method described above.

[0207] The present application also provides a computer-readable storage medium having computer-executable instructions stored therein, which, when executed by a processor, implement the power distribution network electrical risk identification method described above.

[0208] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as SRAM, EEPROM, EPROM, PROM, ROM, magnetic storage, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0209] An exemplary readable storage medium is coupled to the processor, so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an ASIC. Of course, the processor and the readable storage medium can also exist as discrete components in the device.

[0210] The division of units is only a logical functional division, and in actual implementation, there can be another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0211] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e., they can be located in one place or distributed on multiple network units. According to actual needs, some or all of the units can be selected to achieve the purpose of the present embodiment.

[0212] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit.

[0213] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the present application that makes essential contributions to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the embodiments of the method of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a ROM, a RAM, a magnetic disk or an optical disk, and various program code storage media.

[0214] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware. The aforementioned program can be stored in a computer readable storage medium. The program, when executed, executes the steps including the above-mentioned method embodiments; and the aforementioned storage medium includes: a ROM, a RAM, a magnetic disk or an optical disk, and various program code storage media.

[0215] Finally, it should be noted that: those skilled in the art will easily think of other embodiments of the present application after considering the specification and practicing the disclosed application. The present application is intended to cover any variations, uses or adaptations of the present application which follow the general principles of the present application and include known or customary technical means in the art which are not disclosed in the present application, and is not limited to the precise structure described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present application is only limited by the appended claims.

Claims

1. A method for identifying electrical risks in a distribution network, characterized in that: include: Collect electrical parameters, non-electrical parameters and environmental parameters of multiple power devices in the distribution network at multiple times; For each moment, based on the non-electrical parameters of multiple power devices corresponding to the moment, the non-electrical parameters are added to the first knowledge graph to obtain a second knowledge graph corresponding to the moment, where the first knowledge graph includes the connection relationships between the multiple power devices, as well as the geographic locations, device types, and inherent attribute parameters of the multiple power devices; Extracting, from the second knowledge graph corresponding to the moment, attribute features corresponding to the moment and an association between the electrical equipment and the attribute features, the attribute features comprising static attribute features and dynamic attribute features; the dynamic attribute features being extracted from non-electrical parameters in the second knowledge graph; Inputting electrical parameters, environmental parameters, attribute characteristics, and the correlation between electrical equipment and attribute characteristics at multiple moments into a pre-trained electrical risk identification model to predict the electrical risk of each of the multiple electrical equipment; The electrical risk situation is used to indicate whether the electrical equipment has an electrical risk, and the risk type when an electrical risk exists.

2. The method for identifying power-related risks in a distribution network according to claim 1, characterized in that: The electricity-related risk identification model includes a knowledge fusion module, a graph convolutional network, and a Transformer network, wherein: The knowledge fusion module is configured to perform knowledge fusion on the electrical parameters, environmental parameters, and static attribute features to obtain a first fused feature; perform knowledge fusion on the electrical parameters, environmental parameters, and dynamic attribute features to obtain a second fused feature; and concatenate the first fused feature and the second fused feature to obtain a third fused feature; The graph convolutional network is used to perform spatial feature extraction on the third fused feature based on the association between the power equipment and the attribute features to obtain a new representation of the third fused feature; The Transformer network is used to capture the temporal relationship in the new representation of the third fusion feature corresponding to each moment in the multiple moments, and obtain the current power-related risk situation of each power device in the multiple power devices.

3. The method for identifying electrical risks in a distribution network according to claim 1, wherein: Extracting the attribute features corresponding to the moment and the association between the power equipment and the attribute features from the second knowledge graph corresponding to the moment includes: For each electric device, obtaining, from the second knowledge graph corresponding to the moment, a plurality of nodes connected to the electric device and an association relationship between each node and the electric device; the contents of the plurality of nodes including non-electrical parameters associated with the electric device, other electric devices, geographic location, device type, and inherent attribute parameters; Extracting dynamic attribute features corresponding to the moment based on each power device and the nodes to which the power device is connected, whose attributes are non-electrical parameters; and extracting static attribute features corresponding to the moment based on each power device and the nodes to which the device is connected, whose attributes are other power devices, geographic locations, device types, and inherent attribute parameters; According to the association relationship between each node connected to each power device and the power device, the association relationship between the power device and the attribute feature is extracted.

4. The method for identifying electrical risks in a distribution network according to any one of claims 1 to 3, characterized in that: The electricity-related risk identification model is trained in the following way: Construct multiple sets of training sample data sets, each of which includes the electrical parameters, environmental parameters, attribute characteristics of multiple power devices at multiple times, the correlation between power devices and attribute characteristics, and the actual current electrical risk status of each power device; Inputting electrical parameters, environmental parameters, attribute characteristics of multiple power devices at multiple times, and the correlation between the power devices and the attribute characteristics into the power-related risk identification model to be trained, and obtaining the current predicted power-related risk status of each of the multiple power devices; Based on the cross entropy loss function, determine the difference between the predicted electricity-related risk situation and the actual electricity-related risk situation; According to the difference, the model parameters of the electrical risk identification model to be trained are adjusted.

5. The method for identifying electrical risks in a distribution network according to claim 4, characterized in that: The multiple sets of training sample data sets are constructed in the following manner: Acquire multiple sets of historical time series data, each of which includes the electrical parameters, non-electrical parameters, and environmental parameters of multiple power devices in the distribution network at multiple times, as well as the actual current electrical risk status of each power device; For each moment in each set of historical time series data, based on the non-electrical parameters of multiple power devices at the moment in the historical time series data, the non-electrical parameters are added to the first knowledge graph to obtain a second knowledge graph corresponding to the moment; Extracting, from the second knowledge graph corresponding to the moment, the attribute features corresponding to the moment and the association between the power equipment and the attribute features; The multiple sets of training sample data sets are constructed based on the electrical parameters, environmental parameters, attribute characteristics of multiple power equipment in the distribution network at multiple times corresponding to each set of historical time series data, the correlation between the power equipment and the attribute characteristics, and the actual power-related risk conditions of each power equipment.

6. The method for identifying electrical risks in a distribution network according to claim 5, characterized in that: The obtaining of multiple sets of historical time series data includes: Acquire multiple sets of time series data for multiple power devices in the distribution network, each set of time series data includes electrical parameters, non-electrical parameters, and environmental parameters; For each set of time series data, performing data cleaning and normalization processing on the time series data to obtain processed time series data; Based on the timestamps in the processed time series data, performing timestamp alignment processing on the electrical parameters, non-electrical parameters, and environmental parameters in the processed time series data to obtain aligned time series data; Analyze the electrical parameters, non-electrical parameters, and environmental parameters corresponding to each timestamp using expert knowledge to determine the actual electrical risk situation corresponding to the timestamp; The multiple groups of historical time series data are obtained based on each group of aligned time series data and the actual electricity-related risk situation corresponding to each timestamp in the aligned time series data.

7. A device for identifying electrical risks in a power grid, characterized in that: include: A collection unit is used to collect electrical parameters, non-electrical parameters and environmental parameters of multiple power devices in the distribution network at multiple times; a setting unit configured to, for each moment, add the non-electrical parameters of the plurality of electric devices corresponding to the moment to a first knowledge graph, thereby obtaining a second knowledge graph corresponding to the moment, wherein the first knowledge graph includes connection relationships between the plurality of electric devices, as well as geographic locations, device types, and inherent attribute parameters of the plurality of electric devices; an extraction unit, configured to extract, from the second knowledge graph corresponding to the moment, attribute features corresponding to the moment and an association between the electrical equipment and the attribute features, wherein the attribute features include static attribute features and dynamic attribute features; the dynamic attribute features are obtained by extracting non-electrical parameters from the second knowledge graph; A prediction unit is configured to input electrical parameters, environmental parameters, attribute characteristics, and the correlation between electrical equipment and attribute characteristics at multiple moments into a pre-trained electrical risk identification model to predict the electrical risk of each of the multiple electrical equipment; The electrical risk situation is used to indicate whether there is an electrical risk in the power equipment, and the risk type when there is an electrical risk.

8. A computer device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 6 when executed.

10. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 6 when the computer program is executed.

Images