Power failure analysis method and system for smart power grid

Abstract

The embodiment of the invention discloses a power failure analysis method and system for a smart power grid, and relates to the technical field of smart power grids. The method comprises the following steps: carrying out association relation extraction on an association node related to a target node to obtain an association relation embedding vector corresponding to the association node; performing feature extraction on the incidence relation embedding vector to obtain a target relation feature corresponding to the incidence relation embedding vector; determining M associated scenes associated with the target node and the associated node; performing power failure risk assessment according to each associated scene and the corresponding target relation feature to obtain a power failure risk index corresponding to each associated scene; carrying out weight distribution on the association relation embedded vector to obtain an influence factor corresponding to each association scene; and determining a power failure analysis result corresponding to the target node based on the power failure risk index corresponding to the associated scene and the influence factor corresponding to each associated scene.

Description

Technical Field

[0001] The present application relates to the technical field of smart grids, and in particular to a power outage analysis method and system for smart grids. Background Art

[0002] In modern power systems, the development of smart grid technology has made grid operation and management more intelligent and efficient. However, due to various factors, power outages are still inevitable. As an important part of smart grids, power outage analysis undertakes important tasks such as revealing the causes of power outages, predicting power outage risks, and optimizing grid operations.

[0003] Traditional power outage analysis methods mainly rely on manual experience and simple statistical analysis, which are difficult to cope with the increasingly complex power grid environment. Especially when faced with power outages caused by the influence of other nodes, it is difficult to accurately predict the power outage risks of each node.

[0004] Based on this, it is necessary to study a more scientific power outage analysis method and system to improve the accuracy and comprehensiveness of power outage analysis. Summary of the invention

[0005] To solve the above problem, one aspect of an embodiment of this specification provides a power outage analysis method for a smart grid, the method comprising: Extracting association relationships of associated nodes related to the target node to obtain association relationship embedding vectors corresponding to the associated nodes, wherein the associated nodes include adjacent nodes directly connected to the target node and nodes that have an influence relationship with the target node; Performing feature extraction on the association relationship embedding vector to obtain a target relationship feature corresponding to the association relationship embedding vector; Determine M associated scenarios associated with the target node and the associated node, where M is an integer greater than or equal to 1; Perform power outage risk assessment according to each associated scenario and the corresponding target relationship characteristics, and obtain the power outage risk index corresponding to each associated scenario; Performing weight assignment on the association relationship embedding vector to obtain an impact factor corresponding to each of the association scenarios; Based on the power outage risk index corresponding to the associated scenario and the impact factor corresponding to each associated scenario, a power outage analysis result corresponding to the target node is determined, wherein the power outage analysis result at least includes a power outage risk quantification index corresponding to the target node.

[0006] In some embodiments, extracting the association relationship of the associated nodes related to the target node to obtain the association relationship embedding vector corresponding to the associated nodes includes: Obtaining a power grid topology map corresponding to the power grid where the target node is located; Extracting association relationships of associated nodes related to the target node based on the power grid topology map and image recognition technology to obtain association relationship description text corresponding to each associated node; Performing sequence conversion processing on the association relationship description text to obtain an association relationship conversion sequence corresponding to each of the association nodes; The association relationship conversion sequence is embedded and extracted to obtain an association relationship embedding vector corresponding to each association node.

[0007] In some embodiments, performing sequence conversion processing on the association relationship description text to obtain an association relationship conversion sequence corresponding to each of the association nodes includes: After determining the associated nodes related to the target node, obtaining a target processing network for performing power outage analysis on the target node, the target processing network comprising a first processing network; Determine the association relationship description text attribute corresponding to each of the association nodes through the first processing network, and extract the description participle corresponding to the association node based on the association relationship description text attribute; Performing word segmentation semantic mapping on the description word segmentation through the first processing network to obtain word segmentation semantic mapping data corresponding to the description word segmentation; The word segmentation semantic mapping data corresponding to the description word segmentation is serialized through the first processing network to obtain the association relationship conversion sequence of the association node.

[0008] In some embodiments, the target relationship feature is obtained by performing feature mining processing on the association relationship embedding vector through a feature mining network in the first processing network; the first processing network is obtained by: Acquire a first training sample set, wherein the first training sample set includes first sample data without labels; Acquire at least one training task for the first initial processing network, and process the first sample data through the first initial processing network to obtain sample extraction features corresponding to each training task; Determine at least one loss information corresponding to the at least one training task based on the sample extraction feature, wherein each training task corresponds to one loss information; determining first target loss information based on the at least one loss information; Iteratively adjust network parameters of the first initial processing network according to the first target loss information, and use the first initial processing network after parameter adjustment as the first processing network.

[0009] In some embodiments, the target relationship feature is obtained by performing feature mining processing on the association relationship embedding vector through a feature mining network in the first processing network; the target processing network also includes a second processing network connected to the feature mining network; The power outage risk assessment is performed according to each associated scenario and the corresponding target relationship feature to obtain a power outage risk index corresponding to each associated scenario, including: Determining a second processing network from the target processing network, wherein the second processing network includes a dedicated processing network corresponding to each associated scenario; Based on the dedicated processing network corresponding to each associated scenario, a power outage risk assessment is performed on each associated scenario and the corresponding target relationship feature to obtain a power outage risk index corresponding to each associated scenario.

[0010] In some embodiments, the second processing network is obtained based on the following method: Acquire a second training sample set, where the second training sample set includes second sample data annotated with a scene-associated label, where the scene-associated label includes N scene-associated label values ​​corresponding to N associated scenes, respectively, where the N scene-associated label values ​​are determined based on the associated scenes to which the second sample data belongs, and N is an integer greater than or equal to M; Extracting association relationships from the second sample data through the trained first processing network to obtain a first sample association relationship embedding vector for the second sample data; Performing feature extraction processing on the first sample association relationship embedding vector to obtain a first sample feature corresponding to the first sample association relationship embedding vector; Acquire N initial dedicated processing networks corresponding to the N associated scenes respectively, and perform scene type recognition on the first sample features based on the N initial dedicated processing networks respectively, to obtain N scene type recognition probabilities corresponding to the N initial dedicated processing networks; Determining second target loss information based on the N scene type recognition probabilities and the N scene-associated label values; Iteratively adjusting network parameters of the N initial dedicated processing networks based on the second target loss information, and obtaining the second processing network based on the N dedicated processing networks obtained after the parameter adjustment.

[0011] In some embodiments, the association relationship embedding vector is obtained by embedding the association relationship conversion sequence through an embedding layer in a first processing network, and the target processing network further includes a third processing network connected after the embedding layer; The weighting of the association relationship embedding vector is performed to obtain the influence factor corresponding to each of the association scenarios, including: weighting the association relationship embedding vector is performed through the third processing network to obtain the influence factor corresponding to each of the association scenarios.

[0012] In some embodiments, the method further comprises: Acquire a third training sample set, wherein the third training sample set includes third sample data marked with a power outage analysis label; Extracting association relationships from the third sample data through the first processing network to obtain a second sample association relationship embedding vector corresponding to the third sample data; Performing feature extraction processing on the second sample association relationship embedding vector to obtain a second sample feature corresponding to the second sample association relationship embedding vector; Performing a power outage risk assessment on the second sample features through N dedicated processing networks in the second processing network to obtain a sample power outage risk index corresponding to each of the associated scenarios; Performing weight assignment on the second sample association relationship embedding vector through a third initial processing network to obtain a sample influence factor corresponding to each of the association scenarios; Determine a sample power outage analysis result of a sample target node corresponding to the third sample data based on a sample power outage risk index corresponding to each of the associated scenarios and a sample impact factor corresponding to each of the associated scenarios; determining third target loss information based on the power outage analysis tag and the sample power outage analysis result; Iteratively adjust the network parameters of the third initial processing network according to the third target loss information, and obtain the target processing network based on the first processing network, the second processing network and the third processing network after parameter adjustment.

[0013] In some embodiments, determining the power outage analysis result corresponding to the target node based on the power outage risk index corresponding to the associated scenario and the impact factor corresponding to each associated scenario includes: Determining a first power outage analysis result based on the target node's own operating scenario data; Determining a second power outage analysis result corresponding to the target node based on a power outage risk index corresponding to the associated scenario and an impact factor corresponding to each associated scenario; A power outage analysis result corresponding to the target node is obtained according to the first power outage analysis result and the second power outage analysis result.

[0014] Another aspect of the embodiments of this specification also provides a power outage analysis system for a smart grid, the system comprising: A correlation relationship embedding vector generating module, used for extracting correlation relationships from correlation nodes related to a target node to obtain correlation relationship embedding vectors corresponding to the correlation nodes, wherein the correlation nodes include adjacent nodes directly connected to the target node and nodes having an influence relationship with the target node; A target relationship feature determination module is used to extract features from the association relationship embedding vector to obtain a target relationship feature corresponding to the association relationship embedding vector; An associated scene determination module, used to determine M associated scenes associated with the target node and the associated node, where M is an integer greater than or equal to 1; A power outage risk index determination module is used to perform power outage risk assessment according to each associated scenario and the corresponding target relationship characteristics, and obtain a power outage risk index corresponding to each associated scenario; An impact factor determination module, used to assign weights to the association relationship embedding vectors to obtain an impact factor corresponding to each of the association scenarios; A power outage analysis result generation module is used to determine the power outage analysis result corresponding to the target node based on the power outage risk index corresponding to the associated scenario and the impact factor corresponding to each associated scenario, wherein the power outage analysis result at least includes a power outage risk quantification indicator corresponding to the target node.

[0015] The power outage analysis method and system for smart grid provided in the embodiments of this specification may bring at least the following beneficial effects: (1) By considering the power outage risk caused by related nodes, the power outage risk of the target node in a specific scenario can be more comprehensively evaluated, thereby providing more accurate and comprehensive data support for power grid maintenance and emergency response, thereby improving the reliability and risk resistance of power grid operation; (2) By determining the corresponding power outage risk index and impact factor based on the characteristics of the association relationship between each associated node and the target node, the possibility of each associated node causing a power outage on the target node can be accurately evaluated, thereby providing accurate and effective data support for subsequent analysis; (3) The first power outage analysis result is determined based on the target node's own operating scenario data, and the second power outage analysis result corresponding to the target node is determined based on the power outage risk index and impact factor corresponding to the associated node. Finally, the power outage analysis result corresponding to the target node is obtained based on the first power outage analysis result and the second power outage analysis result. The power outage caused by the target node's own factors and the external associated scenarios can be considered at the same time, thereby obtaining a more comprehensive power outage analysis result.

[0016] Additional features will be described in part in the following description. For those skilled in the art, it will become apparent by reviewing the following and accompanying drawings, or it may be understood by the production or operation of the examples. The features of this specification may be implemented and obtained by practicing or using various aspects of the methods, tools, and combinations described in the following detailed examples. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein: Figure 1 is an exemplary flow chart of a power outage analysis method for a smart grid according to some embodiments of this specification; Figure 2 is a flowchart of exemplary sub-steps of a power outage analysis method for a smart grid according to some embodiments of the present specification; Figure 3 is an exemplary power grid topology diagram according to some embodiments of this specification; Figure 4 is an exemplary module diagram of a power outage analysis system for a smart grid according to some embodiments of the present specification. DETAILED DESCRIPTION

[0018] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0019] It should be understood that the "system", "device", "unit" and / or "module" used in this specification is a method for distinguishing different components, elements, parts, parts or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0020] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0021] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0022] The power outage analysis method and system for smart grid provided in the embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0023] Figure 1 is an exemplary flow chart of a power outage analysis method for a smart grid according to some embodiments of the present specification. In some embodiments, the power outage analysis method 100 for a smart grid can be executed by processing logic, which can include hardware (e.g., circuits, dedicated logic, programmable logic, microcode, etc.), software (instructions running on a processing device to perform hardware simulation), etc. or any combination thereof. In some embodiments, Figure 1 One or more operations in the flowchart of the power outage analysis method 100 for smart grids can be implemented by a processing device and / or a terminal device. For example, the power outage analysis method 100 for smart grids can be stored in a storage device in the form of instructions and called and / or executed by a processing device and / or a terminal device.

[0024] Reference Figure 1 In some embodiments, the power outage analysis method 100 for a smart grid may include: Step S110, extracting association relationships from association nodes related to the target node to obtain association relationship embedding vectors corresponding to the association nodes. In some embodiments, step S110 may be performed by the association relationship embedding vector generation module 210 mentioned below.

[0025] There are usually multiple nodes in the power grid, each of which can be understood as a specific power grid device or a part of the power grid. For example, a node can represent a substation, a generator set, a load center, or a connection point of a power grid. In the operation and management of the power grid, these nodes may affect each other. For example, the failure of one node in series may affect the normal operation of other nodes (such as causing power outages in downstream nodes).

[0026] Therefore, in order to accurately analyze the power outage of each node, it is necessary not only to monitor the operation of each node itself, but also to conduct a comprehensive analysis in combination with related associated nodes, which may include adjacent nodes directly connected to the target node and nodes that have an impact relationship with the target node.

[0027] In the embodiment of the present application, in order to realize comprehensive analysis in combination with related associated nodes, it is first necessary to perform step S110 to extract the associated relationships of the associated nodes related to the target node, and obtain the associated relationship embedding vector corresponding to each associated node. Figure 2 In the embodiment of the present application, step S110 may include the following sub-steps: Sub-step S1101, obtaining a power grid topology map corresponding to the power grid where the target node is located.

[0028] It should be noted that in the embodiment of the present application, the target node can be any node in the power grid. For example, the target node can refer to a load node at the user end (for example, a smart meter at the user end can be regarded as a load node at the user end), or a power node in the power grid (for example, a transformer), or a relay node in the transmission network (for example, a distribution box). In the embodiment of the present application, the operating scenario data of each node in the power grid (for example, voltage, current, power factor, temperature, weather conditions, etc.) can be monitored in real time.

[0029] The grid topology map corresponding to the grid where the target node is located can reflect the connection relationship and electrical characteristics between the nodes in the grid. By analyzing the grid topology map, the location of the target node in the grid and its electrical connection relationship with other nodes can be identified. In addition, the grid topology map can also reveal the structural strength and potential weak links of the grid, providing important information for the stable operation and fault analysis of the grid. In an embodiment of the present application, the grid topology map can be obtained by manual drawing or by exporting through a SCADA (Supervisory Control and Data Acquisition) / EMS (Energy Management System) system.

[0030] Sub-step S1102, based on the power grid topology map and image recognition technology, extract the association relationship of the associated nodes related to the target node to obtain the association relationship description text corresponding to each associated node.

[0031] After obtaining the power grid topology map of the power grid where the target node is located, the associated nodes associated with the target node and the specific association relationship may be further determined based on the power grid topology map.

[0032] Specifically, in some embodiments of the present application, the power grid topology map can be processed by image recognition technology to extract association relationships of associated nodes related to the target node, and obtain association relationship description text corresponding to each associated node.

[0033] It can be understood that image recognition technology is a technology that can recognize and analyze image content, and it processes image data by simulating the human visual system. In the embodiment of the present application, image recognition technology may include but is not limited to pattern recognition, machine learning, deep learning and other methods. Through these methods, various graphic elements in the family power grid topology map, such as nodes, lines, equipment identification, etc., can be automatically identified, and the relationship between them can be analyzed, and the corresponding description text can be output. In the embodiment of the present application, a large number of training samples can be used for training to obtain a target image recognition model that can recognize the power grid topology map to extract the associated nodes associated with the target node and output the corresponding description text (for example, the target node and the power grid topology map of the power grid where the target node is located are used as input, and the associated nodes and description text corresponding to the target node are used as output for training). More details about the aforementioned image recognition technology can be regarded as prior art, and will not be discussed in detail in this specification.

[0034] Figure 3 It is an exemplary power grid topology diagram provided in the embodiments of this specification. As an example only, in some embodiments, when the target node is user U3, after the power grid topology diagram is identified by the aforementioned target image recognition model, the associated nodes can be obtained: distribution box P2, distribution box P1, bus M1, distribution transformer T1, where the distribution box can be understood as a distribution node, which can include a distribution cabinet, a distribution station, etc. The description text of the association relationship between distribution box P2 and user U3 is: "Distribution box P2 provides power control services for user U3, and the two are directly connected through distribution lines."; The description text of the association relationship between distribution box P1 and user U3 is: "Distribution box P1 is located in the power path where user U3 is located, and is upstream of distribution box P2 and user U3."; The description text of the association relationship between busbar M1 and user U3 is: "Busbar M1 is the upstream power source of distribution box P2 and distribution box P1, and supplies power through the main distribution line."; The description text of the association relationship between distribution transformer T1 and user U3 is: "Distribution transformer T1 is responsible for converting high voltage electricity into low voltage electricity suitable for user U3, and is the starting point of the entire power supply link.".

[0035] In some embodiments, considering that the associated nodes determined by the image recognition technology may be inaccurate due to problems such as training accuracy, in some embodiments of the present application, the degree of state synchronization between each node and the target node can be determined by analyzing the running scene data of each node, and the associated nodes determined by the image recognition technology can be verified by the degree of synchronization. Specifically, the nodes whose synchronization degree is less than a preset threshold among the associated nodes determined by the image recognition technology and their corresponding associated relationship description texts are deleted, otherwise they are retained.

[0036] Sub-step S1103, performing sequence conversion processing on the association relationship description text to obtain an association relationship conversion sequence corresponding to each of the association nodes.

[0037] In the embodiments of the present application, the sequence conversion processing of the association relationship description text can be understood as abstracting the entities and relationships in the text into a sequence form, and obtaining the association relationship conversion sequence corresponding to each association node, so as to facilitate computer processing. For example, in some embodiments, "distribution box P2 provides power control services for user U3, and the two are directly connected through the distribution line" can be converted into the sequence "P2-U3-distribution line-control", where "P2-U3" represents the direct connection relationship between entities, "distribution line" represents the medium of connection, and "control" represents the nature of the relationship between entities. Similarly, other association relationship description texts can also be converted into corresponding sequences in a similar manner, and they will not be listed one by one here.

[0038] It should be noted that the above sequence conversion process is only an exemplary description. In the embodiment of the present application, other methods can be used to perform sequence conversion on the association relationship description text, which is not specifically limited in this specification.

[0039] In some embodiments of the present application, after determining the associated nodes related to the target node, a target processing network for performing power outage analysis on the target node can be obtained, and the target processing network includes a first processing network. In an embodiment of the present application, the first processing network can be used to determine the associated relationship description text attributes corresponding to each of the associated nodes, and the description segmentation corresponding to the associated node can be extracted based on the associated relationship description text attributes; then, the description segmentation is mapped by the first processing network to obtain the segmentation semantic mapping data corresponding to each of the description segmentation; finally, the segmentation semantic mapping data corresponding to the description segmentation is serialized by the first processing network to obtain the association relationship conversion sequence of the associated node.

[0040] Specifically, the target processing network can be understood as a network structure composed of multiple processing networks (or modules), each of which can be responsible for different processing tasks. Each processing network (or module) can also include one or more sub-modules, which can work together to complete the specified data processing tasks.

[0041] As an example only, in some embodiments of the present application, the first processing network may include: an entity recognition submodule, used to identify the attributes of the aforementioned association relationship description text, so as to obtain the entities therein, such as distribution boxes, users, distribution lines, etc.; a relationship extraction submodule, used to extract the relationships between entities from the association relationship description text, such as providing services, direct connections, etc.; a serialization submodule, used to convert the identified entities and relationships into a sequence form for subsequent processing; a word segmentation semantic mapping submodule, used to map the description word segmentation to the corresponding semantic information to enhance the semantic expression ability of the sequence; a sequence conversion submodule, used to convert the mapped word segmentation sequence into the final association relationship conversion sequence.

[0042] Through the collaborative work of these submodules, the association relationship description text can be effectively converted into computer-processable sequence data, and then the association relationship characteristics between each associated node and the target node can be obtained for subsequent power outage analysis and processing tasks. It can be understood that in actual applications, these submodules can be adjusted and optimized according to specific needs to improve the accuracy and efficiency of processing.

[0043] In some embodiments of the present application, the first processing network is obtained by: First, obtain a first training sample set, which includes first sample data without a label. In an embodiment of the present application, the first sample data can be text sample data describing an association relationship. Secondly, obtain at least one training task for the first initial processing network, and process the first sample data through the first initial processing network to obtain sample extraction features corresponding to each training task. Then, based on the sample extraction features, determine at least one loss information corresponding to the at least one training task, wherein each training task corresponds to a loss information. Further, determine the first target loss information based on the at least one loss information. Finally, iteratively adjust the network parameters of the first initial processing network according to the first target loss information, and use the first initial processing network after parameter adjustment as the first processing network.

[0044] It should be noted that, in the embodiment of the present application, the aforementioned loss information can be used to measure the difference between the prediction results of each submodule and the real data. The loss information for each training task can be determined based on the difference between the processing result determined by the corresponding sample extraction feature and the training target of the corresponding training task. The first target loss information can be loss information for training and evaluating the first initial processing network determined based on at least one loss information. The first target loss information can be obtained by averaging at least one loss information, or by weighted summing at least one loss information. The weight corresponding to each loss information can be determined according to actual needs, and is not specifically limited in this specification.

[0045] As an example only, in some embodiments of the present application, the aforementioned loss information can be calculated by reconstruction loss, contrast loss, etc. In the embodiment of the present application, the network parameters of the first initial processing network can be iteratively adjusted according to the first target loss information, and the aforementioned first processing network can be obtained based on the first initial processing network after parameter adjustment.

[0046] Sub-step S1104, performing embedded association extraction on the association conversion sequence to obtain an association embedding vector corresponding to each of the association nodes.

[0047] In an embodiment of the present application, embedding the aforementioned association relationship conversion sequence for association relationship extraction can be understood as mapping the entities and relationships in each sequence into a high-dimensional space, generating corresponding numerical vectors, and thus obtaining the association relationship embedding vector corresponding to each associated node. For example, in some embodiments, a pre-trained word embedding model, such as Word2Vec or BERT, can be used to convert the entities and relationships in the aforementioned association relationship conversion sequence into numerical vectors, and capture the semantic information between the entities and relationships for subsequent calculation and analysis. During the conversion process, each entity and relationship can be assigned a vector, and the dimensions and values ​​of these vectors can be determined by the pre-trained model. In some embodiments, when there are two or more groups of entity relationships in the association relationship conversion sequence, a multi-dimensional association relationship embedding vector can be generated. In an embodiment of the present application, in this way, the association relationship in text form can be converted into a numerical form that can be processed by a computer, thereby providing data support for subsequent analysis.

[0048] It should be understood that the above implementation method of extracting the association relationship of the associated nodes related to the target node and obtaining the association relationship embedding vector corresponding to the associated node is only an exemplary description. In some other embodiments of the present application, other methods can be used to obtain the associated nodes related to the target node and the association relationship embedding vector corresponding to each associated node. For example, in some embodiments, the degree of state synchronization between each node and the target node (such as power-on state, voltage fluctuation state, current fluctuation state, etc.) can be determined by analyzing the operating scenario data of each node, and then the strength of the association between them can be determined (the higher the degree of synchronization, the higher the strength of the association), and then the nodes with an association strength greater than a preset threshold are selected as the associated nodes related to the target node, and the corresponding association relationship embedding vector is generated according to the association strength.

[0049] Step S120, extracting features from the association relationship embedding vector to obtain target relationship features corresponding to the association relationship embedding vector. In some embodiments, step S120 may be performed by the target relationship feature determination module 220 mentioned below.

[0050] After obtaining the association relationship embedding vector corresponding to each associated node associated with the target node through the above steps, the association relationship embedding vector can be further feature extracted to obtain the target relationship feature corresponding to the association relationship embedding vector. In the embodiment of the present application, the process of feature extraction of the association relationship embedding vector can be regarded as an information filtering process, the purpose of which is to remove redundant information and invalid information, retain key features, so as to improve the accuracy and efficiency of subsequent processing.

[0051] Specifically, since the association relationship embedding vector obtained in the above process may contain some invalid relationship information irrelevant to the power outage analysis or some redundant information of repeated records, for example, in some embodiments, the aforementioned association relationship embedding vector may contain information of some other nodes in parallel with the target node. Since it has no effect on the subsequent power outage analysis, it can be regarded as invalid information and removed.

[0052] In some embodiments of the present application, when the aforementioned association relationship embedding vector is a multidimensional vector, the process of extracting features from the association relationship embedding vector can be implemented through an attention mechanism. Specifically, the attention mechanism can assign different weights to different parts of the association relationship embedding vector, thereby highlighting important features and suppressing unimportant features, thereby extracting features from the aforementioned association relationship embedding vector and obtaining the target relationship features corresponding to the association relationship embedding vector.

[0053] It can be understood that in the embodiment of the present application, after performing feature extraction on the association relationship embedding vector, a more refined target relationship feature vector can be obtained, and the target relationship feature vector can more accurately reflect the relationship between the target node and other associated nodes.

[0054] In some embodiments of the present application, the aforementioned first processing network may further include a feature mining network, and the aforementioned target relationship features may be obtained by processing the feature mining network. It should be noted that in the embodiments of the present application, the feature mining network may be a model based on the Transformer structure (converter model). It can be understood that Transformer is a deep learning model based on the self-attention mechanism, and its essence is an encoder-decoder structure. For example, the feature mining network in the first processing network may be a Transformer model with 4 depth and 8 heads, wherein 4 depth means that the depth of the Transformer model is 4, that is, the number of encoders is 4, and 8 heads means that the number of heads of the multi-head attention module in the Transformer model is 4, that is, the number of times the multi-head attention module can repeat the calculation in parallel is 4. As a result, the first processing network can be used to extract features that accurately characterize the relationship between the target node and the associated nodes (that is, the target relationship features), thereby improving the accuracy of subsequent power outage analysis. Step S130, determining M associated scenarios associated with the target node and the associated node. In some embodiments, step S130 may be performed by the associated scenario determination module 230 mentioned below.

[0055] In an embodiment of the present application, the associated scenario can be understood as an operating scenario of an associated node that has an associated relationship with the target node. Each operating scenario may include a set of operating environment data corresponding to an associated node, such as historical power outage records, real-time operating data (such as voltage, current, load, etc.), temperature, humidity, weather conditions, etc.

[0056] Step S140, perform power outage risk assessment according to each associated scenario and the corresponding target relationship feature, and obtain a power outage risk index corresponding to each associated scenario. In some embodiments, step S140 may be performed by the power outage risk index determination module 240 mentioned below.

[0057] In an embodiment of the present application, a second processing network can be cascaded after the feature mining network of the first processing network, and the second processing network can include N dedicated processing networks (N≥M) corresponding to N associated scenarios, wherein each associated scenario can correspond to a dedicated processing network. In other words, in an embodiment of the present application, each dedicated processing network can be used to process an associated scenario and its corresponding target relationship features, thereby performing a power outage risk assessment on each associated node respectively, and obtaining a power outage risk index corresponding to each associated scenario.

[0058] In the embodiment of the present application, the second processing network can be obtained based on the following method: First, obtain a second training sample set, the second training sample set includes second sample data annotated with a scene association label, wherein the second sample data may include a number of association relationship description texts and their corresponding associated scene data (i.e., each second sample data may be a set of operating environment data corresponding to an associated node), the scene association label includes N scene association label values ​​corresponding to N associated scenes, the N scene association label values ​​are determined based on the associated scene to which the second sample data belongs (the association label value can be used to characterize the power outage risk index of the corresponding associated scene, which can be obtained through expert experience evaluation), and N is an integer greater than or equal to M. In an embodiment of the present application, N may represent the number of dedicated processing networks included in the second processing network.

[0059] Secondly, the second sample data is subjected to association extraction through the trained first processing network to obtain a first sample association embedding vector of the second sample data, and feature extraction processing is performed on the first sample association embedding vector to obtain a first sample feature corresponding to the first sample association embedding vector.

[0060] Furthermore, N initial dedicated processing networks (which may be a classifier in nature) corresponding to the N associated scenarios are obtained, and based on the N initial dedicated processing networks, the scene type identification is performed on the first sample features respectively, to obtain N scene type identification probabilities corresponding to the N initial dedicated processing networks. It should be noted that in an embodiment of the present application, each associated scenario may be processed by any one of the N initial dedicated processing networks, or may be processed separately by the N initial dedicated processing networks. The scene type identification probability may be used to characterize the power outage risk index reflected by the corresponding operating environment data. When an associated scenario is processed separately by N initial dedicated processing networks, the scene type identification probability may refer to the average of the N power outage risk indices obtained by processing the N initial dedicated processing networks.

[0061] Then, the second target loss information is determined based on the N scene type recognition probabilities and the N scene associated label values. In the embodiment of the present application, the second target loss information can be calculated by the mean square error loss, and its specific calculation method can be regarded as the prior art, and will not be discussed in detail in this specification.

[0062] Finally, the network parameters of the N initial dedicated processing networks are iteratively adjusted based on the second target loss information, and the second processing network is obtained based on the N dedicated processing networks obtained after the parameter adjustment.

[0063] It can be understood that in an embodiment of the present application, the above-mentioned second processing network can be a deep learning model, which can perform power outage risk assessment on each associated node by learning the relationship between associated scenarios and target relationship features. After the second processing network is trained through the above steps, the power outage risk assessment can be performed according to each associated scenario and the corresponding target relationship feature, and the power outage risk index corresponding to each associated scenario can be obtained. Step S150, weights are assigned to the associated relationship embedding vector to obtain the impact factor corresponding to each associated scenario. In some embodiments, step S150 can be performed by the impact factor determination module 250 mentioned later.

[0064] In an embodiment of the present application, the association relationship embedding vector can be obtained by embedding the association relationship conversion sequence through the embedding layer in the first processing network. The target processing network also includes a third processing network connected after the embedding layer. In an embodiment of the present application, the association relationship embedding vector can be weighted by the third processing network to obtain the impact factor corresponding to each of the association scenarios. It can be understood that the impact factor can be used to characterize the contribution of each association scenario to the final power outage analysis result.

[0065] In some embodiments, the third processing network may include a Transformer model, a fully connected layer, and a softmax layer, wherein the Transformer model can be used to perform feature extraction processing on the association relationship embedding vector, thereby obtaining feature information that can be used for weight extraction, and the fully connected layer and the softmax layer can be used to determine the impact factors corresponding to each associated scenario based on the aforementioned feature information. Specifically, the output of the Transformer model can be further processed by the fully connected layer to enhance the model's ability to distinguish the importance of different associated scenarios. The softmax layer converts the output of the fully connected layer into a probability distribution to ensure that the impact factors of each associated scenario add up to 1, thereby reflecting the relative importance of each associated scenario in the power outage analysis.

[0066] In the embodiment of the present application, the aforementioned association relationship embedding vector may carry the distance information and connection relationship information between the association node and the target node, and the impact factor corresponding to each association scenario may be determined based on the distance information and connection relationship information. As an example only, in some embodiments, the impact factor may be obtained based on the following formula: Where E represents the impact factor, Represents a function, which can be a linear or nonlinear function, used to calculate the impact factor of the associated scene. For example, the impact factor can be calculated using a weighted sum method, where the distance information and connection relationship information can be used as a weighting factor (in the embodiment of the present application, the aforementioned connection relationship information can be obtained through data mapping processing ). In some embodiments, a machine learning method, such as a support vector machine (SVM) or a neural network, may be used to learn the parameters of this function, thereby accurately reflecting the impact of different association scenarios on the power outage analysis results.

[0067] In some embodiments of the present application, a third training sample set may be obtained. In the embodiments of the present application, the third training sample set includes third sample data marked with a power outage analysis label, and the third sample data may include a number of associated nodes and their corresponding associated scenarios (i.e., operating environment data). Then, the third sample data is subjected to association extraction through the aforementioned first processing network to obtain a second sample association embedding vector corresponding to the third sample.

[0068] Furthermore, by performing feature extraction processing on the second sample association relationship embedding vector, a second sample feature corresponding to the second sample association relationship embedding vector can be obtained; by performing power outage risk assessment on the second sample feature through N dedicated processing networks in the second processing network, a sample power outage risk index corresponding to each of the associated scenarios can be obtained.

[0069] Furthermore, the second sample association relationship embedding vector is weighted by the third initial processing network to obtain a sample impact factor corresponding to each of the associated scenarios. Based on the sample power outage risk index corresponding to each of the associated scenarios and the sample impact factor corresponding to each of the associated scenarios, the sample power outage analysis result of the sample target node corresponding to the third sample data is determined. In an embodiment of the present application, the power outage analysis result includes at least a power outage risk quantification index corresponding to the target node, and the calculation process of the power outage risk quantification index can be expressed as follows: in, It represents the quantitative index of the power outage risk corresponding to the target node m, which represents the power outage risk probability corresponding to the target node m. represents the power outage risk index corresponding to the i-th associated node related to the target node; represents the impact factor corresponding to the i-th associated node related to the target node; M represents the number of associated nodes related to the target node; Represents a normalized calculation. Further, the third target loss information is determined based on the power outage analysis label and the sample power outage analysis result. In an embodiment of the present application, the power outage analysis label can reflect the power outage analysis results corresponding to several associated nodes in the third sample data (the power outage analysis results can be understood as the possibility that the several associated nodes cause a power outage at the target node). In an embodiment of the present application, the third target loss information can be obtained based on the difference between the power outage analysis label and the power outage analysis result obtained by the aforementioned process. Specifically, in an embodiment of the present application, the third target loss information can be obtained based on the power outage risk quantification index in the power outage analysis label and the power outage analysis result. In the embodiment of the present application, the network parameters of the third initial processing network are iteratively adjusted according to the third target loss information, and the target processing network is obtained based on the first processing network, the second processing network and the third processing network after parameter adjustment obtained by the aforementioned process.

[0070] It should be pointed out that in the embodiments of the present application, the iterative adjustment process of each of the above-mentioned processing networks can be carried out for the purpose of minimizing losses, which can be regarded as prior art, and the specific adjustment process thereof will no longer be described in this specification. In addition, it should be noted that after the aforementioned first processing network completes training, its parameters can be fixed, and in the subsequent process, only the parameters of the second processing network can be adjusted. Similarly, after the second processing network completes training, the parameters of the second processing network can be fixed, and in the subsequent process, only the parameters of the third processing network can be adjusted. Step S160, based on the power outage risk index corresponding to the associated scenario and the influencing factor corresponding to each associated scenario, determine the power outage analysis result corresponding to the target node. In some embodiments, step S160 can be performed by the power outage analysis result generation module 260 mentioned later.

[0071] From the above content, it can be known that in the embodiment of the present application, the power outage risk index corresponding to each associated scenario related to the target node and the impact factor corresponding to each associated scenario can be obtained through the target processing network. After obtaining the power outage risk index corresponding to each associated scenario and its corresponding impact factor, it can be based on the above formula Calculate and obtain the power outage analysis results corresponding to the target node.

[0072] It should be noted that the above power outage analysis results are the probabilities that each associated node may cause a power outage at the target node, calculated only based on the operating environment data corresponding to the associated nodes. It can be used to evaluate the probability of a power outage at the target node, but does not take into account power outages caused by its own factors.

[0073] Based on this, in some embodiments of the present application, a first power outage analysis result can be determined based on the target node's own operating scenario data, and a second power outage analysis result corresponding to the target node can be determined based on the power outage risk index corresponding to the associated scenario and the influencing factor corresponding to each associated scenario. Finally, the power outage analysis result corresponding to the target node is obtained based on the first power outage analysis result and the second power outage analysis result. Among them, the first power outage analysis result can be obtained by processing a trained machine learning model. For example, a machine learning model can be used to analyze the target node's own operating environment data to predict the risk of power outages that may be caused by the node's own factors. The model can be a support vector machine, a neural network, or other types of prediction models, which will learn the complex relationship between node operating data and power outage events during the training process. As an example only, the machine learning model can be trained in the following way: S1. Obtain sample data and its corresponding label, wherein the sample data includes sample operating environment data of the sample target node, such as voltage, current, load, temperature, humidity, weather conditions, historical power outage records and other parameters. The label can be used to indicate whether a power outage event occurs at the sample target node under the aforementioned sample operating environment data.

[0074] S2, taking the sample operating environment data as input and the label corresponding to the sample operation data as output to train the initial machine learning model, and terminating the training when the preset conditions are met (for example, the loss function reaches a preset threshold or the number of iterations reaches a preset value), to obtain a trained machine learning model.

[0075] S3, use the validation set to validate the machine learning model trained in the previous steps to ensure that the model has good generalization ability.

[0076] More details about model training can be regarded as prior art and will not be repeated in this specification. It should be understood that after the trained machine learning model is obtained through the above steps, the trained machine learning model can be used to process the target node's own operating scenario data to determine the above first power outage analysis result.

[0077] The second power outage analysis result can be calculated based on the method described above, which will not be described in detail here.

[0078] In an embodiment of the present application, the power outage analysis result corresponding to the target node can be obtained according to the first power outage analysis result and the second power outage analysis result. As an example only, in some embodiments, the power outage analysis result corresponding to the target node can be obtained by performing a weighted operation on the first power outage analysis result and the second power outage analysis result. The process can be expressed as follows:

[0079] Where T represents the power outage analysis result corresponding to the target node (which represents the possibility of a power outage event occurring in the target phase). Indicates the first power outage analysis result, Indicates the second power outage analysis result, represents the weight coefficient corresponding to the first power outage analysis result, Indicates the weight coefficient corresponding to the second power outage analysis result. In the embodiment of the present application, Can be greater than ( and are all between 0 and 1) to highlight the weight of the first power outage analysis result in the calculation process. In some embodiments, can be equal to 1, It can be between 0 and 0.5. At this time, if If it is greater than 1, the power outage analysis result T corresponding to the target node finally calculated is 1. It can be understood that in the embodiment of the present application, through the above method, the influence of the target node's own factors and the influence of external related scenarios can be considered at the same time, so as to obtain a more comprehensive power outage analysis result.

[0080] Figure 4 is an exemplary module diagram of a power outage analysis system for a smart grid provided according to some embodiments of this specification. Figure 4 In an embodiment of the present application, a power outage analysis system 200 for a smart grid may include an association relationship embedding vector generation module 210, a target relationship feature determination module 220, an association scenario determination module 230, a power outage risk index determination module 240, an impact factor determination module 250, and a power outage analysis result generation module 260.

[0081] The association relationship embedding vector generating module 210 may be used to extract association relationships from association nodes related to the target node, and obtain association relationship embedding vectors corresponding to the association nodes.

[0082] The target relationship feature determination module 220 may be used to perform feature extraction on the association relationship embedding vector to obtain the target relationship feature corresponding to the association relationship embedding vector.

[0083] The associated scene determination module 230 may be configured to determine M associated scenes associated with the target node and the associated node.

[0084] The power outage risk index determination module 240 may be used to perform power outage risk assessment according to each associated scenario and the corresponding target relationship feature, and obtain a power outage risk index corresponding to each associated scenario.

[0085] The impact factor determination module 250 may be used to assign weights to the association relationship embedding vectors to obtain an impact factor corresponding to each of the association scenarios.

[0086] The power outage analysis result generating module 260 may be configured to determine a power outage analysis result corresponding to the target node based on a power outage risk index corresponding to the associated scenario and an impact factor corresponding to each associated scenario.

[0087] For more details about the above modules, please refer to other places in this manual (for example Figure 1~Figure 3 part and its related description) will not be repeated here.

[0088] It should be understood that Figure 4 The power outage analysis system 200 for smart grids and its modules shown can be implemented in various ways. For example, in some embodiments, the system and its modules can be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art will understand that the above methods and systems can be implemented using computer executable instructions and / or included in processor control codes, such as such codes provided on carrier media such as disks, CDs or DVD-ROMs, programmable memories such as read-only memories (firmware), or data carriers such as optical or electronic signal carriers. The system and its modules of this specification can be implemented not only by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., but can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above hardware circuits and software (e.g., firmware).

[0089] It should be noted that the above description of the power outage analysis system 200 for smart grid is provided for illustrative purposes only and is not intended to limit the scope of this specification. It is understandable that for those skilled in the art, according to the description of this specification, without departing from this principle, the modules can be arbitrarily combined, or a subsystem can be formed to connect with other modules. For example, Figure 4The association relationship embedding vector generation module 210, target relationship feature determination module 220, association scenario determination module 230, power outage risk index determination module 240, impact factor determination module 250, and power outage analysis result generation module 260 described in the embodiment can be different modules in one system, or one module can realize the functions of two or more modules mentioned above. Such variations are all within the protection scope of this specification.

[0090] In summary, the beneficial effects that may be brought about by the embodiments of this specification include but are not limited to: (1) In the power outage analysis method and system for smart grid provided in some embodiments of this specification, by considering the power outage risk caused by related nodes, the power outage risk of the target node in a specific scenario can be more comprehensively evaluated, thereby providing more accurate and comprehensive data support for the maintenance and emergency response of the power grid, thereby achieving the effect of improving the reliability and risk resistance of the power grid operation; (2) In the power outage analysis method and system for smart grid provided in some embodiments of this specification, by determining the corresponding power outage risk index and influencing factor according to the characteristics of the association relationship between each associated node and the target node, it can In order to accurately evaluate the possibility of each associated node causing a power outage on the target node, thereby providing accurate and effective data support for subsequent analysis; (3) In the power outage analysis method and system for smart grid provided in some embodiments of this specification, a first power outage analysis result is determined based on the target node's own operating scenario data, and a second power outage analysis result corresponding to the target node is determined based on the power outage risk index and impact factor corresponding to the associated node. Finally, the power outage analysis result corresponding to the target node is obtained based on the first power outage analysis result and the second power outage analysis result. The power outage caused by the target node's own factors and external associated scenarios can be considered at the same time, thereby obtaining a more comprehensive power outage analysis result.

[0091] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced may be any one or a combination of the above, or may be any other beneficial effects that may be obtained.

[0092] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0093] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0094] In addition, it will be understood by those skilled in the art that various aspects of this specification may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of this specification may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of this specification may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0095] The computer storage medium may contain a propagation data signal containing computer program code, for example, on baseband or as part of a carrier. The propagation signal may have a variety of manifestations, including electromagnetic form, optical form, etc., or a suitable combination. The computer storage medium may be any computer-readable medium other than a computer-readable storage medium, which can be connected to an instruction execution system, device or equipment to achieve communication, propagation or transmission of programs for use. The program code located on the computer storage medium can be propagated through any suitable medium, including radio, cable, fiber optic cable, RF, or similar media, or any combination of the above media. The computer program code required for the operation of each part of this specification can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, etc., conventional procedural programming languages ​​such as C language, Visual Basic, Fortran2003, Perl, COBOL2002, PHP, ABAP, dynamic programming languages ​​such as Python, Ruby and Groovy, or other programming languages. The program code may run entirely on the user's computer, or as a stand-alone software package, or partly on the user's computer and partly on a remote computer, or entirely on a remote computer or processing device. In the latter case, the remote computer may be connected to the user's computer via any form of network, such as a local area network (LAN) or wide area network (WAN), or to an external computer (e.g., via the Internet), or in a cloud computing environment, or as a service such as software as a service (SaaS).

[0096] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing processing device or mobile device.

[0097] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0098] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining the digits. Although the numerical domains and parameters used to confirm the breadth of the range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0099] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification is hereby incorporated by reference in its entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0100] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A power outage analysis method for a smart grid, characterized in that: The method comprises: Extracting association relationships of associated nodes related to the target node to obtain association relationship embedding vectors corresponding to the associated nodes, wherein the associated nodes include adjacent nodes directly connected to the target node and nodes that have an influence relationship with the target node; Performing feature extraction on the association relationship embedding vector to obtain a target relationship feature corresponding to the association relationship embedding vector; Determine M associated scenarios associated with the target node and the associated node, where M is an integer greater than or equal to 1; Perform power outage risk assessment according to each associated scenario and the corresponding target relationship characteristics, and obtain the power outage risk index corresponding to each associated scenario; Performing weight assignment on the association relationship embedding vector to obtain an impact factor corresponding to each of the association scenarios; Based on the power outage risk index corresponding to the associated scenario and the impact factor corresponding to each associated scenario, a power outage analysis result corresponding to the target node is determined, wherein the power outage analysis result at least includes a power outage risk quantification index corresponding to the target node.

2. The method according to claim 1, characterized in that The extracting the association relationship of the associated nodes related to the target node to obtain the association relationship embedding vector corresponding to the associated nodes includes: Obtaining a power grid topology map corresponding to the power grid where the target node is located; Extracting association relationships of associated nodes related to the target node based on the power grid topology map and image recognition technology to obtain association relationship description text corresponding to each associated node; Performing sequence conversion processing on the association relationship description text to obtain an association relationship conversion sequence corresponding to each of the association nodes; The association relationship conversion sequence is embedded and extracted to obtain an association relationship embedding vector corresponding to each association node.

3. The method according to claim 2, characterized in that The performing sequence conversion processing on the association relationship description text to obtain an association relationship conversion sequence corresponding to each association node includes: After determining the associated nodes related to the target node, obtaining a target processing network for performing power outage analysis on the target node, the target processing network comprising a first processing network; Determining the association relationship description text attribute corresponding to each of the association nodes through the first processing network, and extracting the description participle corresponding to the association node based on the association relationship description text attribute; Performing word segmentation semantic mapping on the description word segmentation through the first processing network to obtain word segmentation semantic mapping data corresponding to the description word segmentation; The word segmentation semantic mapping data corresponding to the description word segmentation is serialized through the first processing network to obtain the association relationship conversion sequence of the association node.

4. The method according to claim 3, characterized in that The target relationship feature is obtained by performing feature mining processing on the association relationship embedding vector by a feature mining network in the first processing network; the first processing network is obtained by: Acquire a first training sample set, wherein the first training sample set includes first sample data without labels; Acquire at least one training task for the first initial processing network, and process the first sample data through the first initial processing network to obtain sample extraction features corresponding to each training task; Determine at least one loss information corresponding to the at least one training task based on the sample extraction feature, wherein each training task corresponds to one loss information; determining first target loss information based on the at least one loss information; Iteratively adjust network parameters of the first initial processing network according to the first target loss information, and use the first initial processing network after parameter adjustment as the first processing network.

5. The method according to claim 3, characterized in that The target relationship feature is obtained by performing feature mining processing on the association relationship embedding vector through a feature mining network in the first processing network; the target processing network also includes a second processing network connected to the feature mining network; The power outage risk assessment is performed according to each associated scenario and the corresponding target relationship feature to obtain the power outage risk index corresponding to each associated scenario, including: Determining a second processing network from the target processing network, wherein the second processing network includes a dedicated processing network corresponding to each associated scenario; Based on the dedicated processing network corresponding to each associated scenario, a power outage risk assessment is performed on each associated scenario and the corresponding target relationship feature to obtain a power outage risk index corresponding to each associated scenario.

6. The method according to claim 5, characterized in that The second processing network is obtained based on the following method: Acquire a second training sample set, where the second training sample set includes second sample data annotated with a scene-associated label, where the scene-associated label includes N scene-associated label values ​​corresponding to N associated scenes, respectively, where the N scene-associated label values ​​are determined based on the associated scenes to which the second sample data belongs, and N is an integer greater than or equal to M; Extracting association relationships from the second sample data through the trained first processing network to obtain a first sample association relationship embedding vector for the second sample data; Performing feature extraction processing on the first sample association relationship embedding vector to obtain a first sample feature corresponding to the first sample association relationship embedding vector; Acquire N initial dedicated processing networks corresponding to the N associated scenes respectively, and perform scene type recognition on the first sample features based on the N initial dedicated processing networks respectively, to obtain N scene type recognition probabilities corresponding to the N initial dedicated processing networks; Determining second target loss information based on the N scene type recognition probabilities and the N scene-associated label values; Iteratively adjusting network parameters of the N initial dedicated processing networks based on the second target loss information, and obtaining the second processing network based on the N dedicated processing networks obtained after the parameter adjustment.

7. The method according to claim 6, characterized in that The association relationship embedding vector is obtained by embedding the association relationship conversion sequence through an embedding layer in a first processing network, and the target processing network also includes a third processing network connected after the embedding layer; The weighting of the association relationship embedding vector is performed to obtain the influence factor corresponding to each of the association scenarios, including: weighting the association relationship embedding vector is performed through the third processing network to obtain the influence factor corresponding to each of the association scenarios.

8. The method according to claim 7, characterized in that The method further comprises: Acquire a third training sample set, wherein the third training sample set includes third sample data marked with a power outage analysis label; Extracting association relationships from the third sample data through the first processing network to obtain a second sample association relationship embedding vector corresponding to the third sample data; Performing feature extraction processing on the second sample association relationship embedding vector to obtain a second sample feature corresponding to the second sample association relationship embedding vector; Performing a power outage risk assessment on the second sample features through N dedicated processing networks in the second processing network to obtain a sample power outage risk index corresponding to each of the associated scenarios; Performing weight assignment on the second sample association relationship embedding vector through a third initial processing network to obtain a sample influence factor corresponding to each of the association scenarios; Determine a sample power outage analysis result of a sample target node corresponding to the third sample data based on a sample power outage risk index corresponding to each of the associated scenarios and a sample impact factor corresponding to each of the associated scenarios; determining third target loss information based on the power outage analysis tag and the sample power outage analysis result; Iteratively adjust the network parameters of the third initial processing network according to the third target loss information, and obtain the target processing network based on the first processing network, the second processing network and the third processing network after parameter adjustment.

9. The method according to any one of claims 1 to 8, characterized in that The determining, based on the power outage risk index corresponding to the associated scenario and the impact factor corresponding to each associated scenario, a power outage analysis result corresponding to the target node includes: Determining a first power outage analysis result based on the target node's own operating scenario data; Determining a second power outage analysis result corresponding to the target node based on a power outage risk index corresponding to the associated scenario and an impact factor corresponding to each associated scenario; A power outage analysis result corresponding to the target node is obtained according to the first power outage analysis result and the second power outage analysis result.

10. A power outage analysis system for a smart grid, characterized in that: The system comprises: A correlation relationship embedding vector generating module, used for extracting correlation relationships from correlation nodes related to a target node to obtain correlation relationship embedding vectors corresponding to the correlation nodes, wherein the correlation nodes include adjacent nodes directly connected to the target node and nodes having an influence relationship with the target node; A target relationship feature determination module is used to extract features from the association relationship embedding vector to obtain a target relationship feature corresponding to the association relationship embedding vector; An associated scene determination module, used to determine M associated scenes associated with the target node and the associated node, where M is an integer greater than or equal to 1; A power outage risk index determination module is used to perform power outage risk assessment according to each associated scenario and the corresponding target relationship characteristics, and obtain a power outage risk index corresponding to each associated scenario; An impact factor determination module, used to assign weights to the association relationship embedding vectors to obtain an impact factor corresponding to each of the association scenarios; A power outage analysis result generation module is used to determine the power outage analysis result corresponding to the target node based on the power outage risk index corresponding to the associated scenario and the impact factor corresponding to each associated scenario, wherein the power outage analysis result at least includes a power outage risk quantification indicator corresponding to the target node.

Images