Method and System for Identifying Hidden Dangers of Cable Discharge in Distribution Network Based on Cloud-Edge Collaboration

Abstract

The present application discloses a method and system for identifying hidden dangers of cable discharge in a cloud-edge collaborative distribution network, including: obtaining the severity level, cause of the hidden danger, and occurrence location of the hidden danger event; extracting semantic features of the cause of the hidden danger and classifying the hidden danger event into multiple risk types; thereby obtaining a triple dataset composed of severity level - risk type - environmental attribute; based on the environmental attribute of the monitoring area, determining the risk types associated with the monitoring area according to the environmental attribute combination strongly associated with the risk type, and determining the weights of the risk types; the edge server configures an identification model according to the risk types in each monitoring area, and configures identification resources for the identification tasks corresponding to the risk types according to the weights of the risk types in the monitoring area, so as to identify various types of discharge hidden dangers, wherein, the risk types with higher weights are allocated more identification resources than those with lower weights.

Description

Technical Field

[0001] This application relates to edge computing and artificial intelligence, and particularly to a method and system for identifying hidden dangers of cable discharge in a distribution network with cloud-edge collaboration. Background Art

[0002] Among related technologies, due to the widespread existence of hidden danger discharge phenomena, such as lightning strikes, icing, tree obstacles, pollution flashovers, foreign object hanging, etc., these hidden dangers are common in the distribution network. During the operation of the power grid, it is necessary to regularly identify these risks, and the occurrence of these hidden dangers usually has certain environmental feature correlations with the environment where the distribution network is located. Therefore, some mathematical models or AI models can be used to identify a certain type of hidden danger of discharge. In order to identify different hidden dangers, various types of models have been designed and trained in the prior art. However, the types of hidden dangers that appear vary in different environments. Therefore, if all the identification models are configured for the monitoring area and all perform identification and analysis at the same frequency, the efficiency is not high. Summary of the Invention

[0003] The present invention aims to at least solve one of the technical problems existing in the prior art. For this purpose, the present invention provides a method and system for identifying hidden dangers of cable discharge in a distribution network with cloud-edge collaboration to optimize the identification efficiency of hidden danger discharge of cables during the operation of the distribution network.

[0004] On the one hand, an embodiment of the present application provides a method for identifying hidden dangers of cable discharge in a distribution network with cloud-edge collaboration, including the following steps:

[0005] Obtain the severity level, cause of the hidden danger event, and occurrence location;

[0006] According to the cause of the hidden danger, classify the hidden danger event into multiple risk types; specifically, extract semantic features from the cause of the hidden danger, cluster the feature vectors after semantic feature extraction, perform sampling analysis on the clustering results to obtain keywords, and screen the clustering results based on the keywords to classify the hidden danger event into multiple risk types;

[0007] Based on the map information, determine the environmental attributes of the occurrence location, so as to obtain a triple dataset composed of severity level - risk type - environmental attributes; through statistical analysis of the triple dataset, determine the environmental attribute combinations strongly associated with the risk type;

[0008] Based on the environmental attributes of the monitoring area, determine the risk types associated with the monitoring area according to the environmental attribute combinations strongly associated with the risk type, and determine the weights of the risk types;

[0009] Allocate each monitoring area to an edge server, and one edge server serves multiple monitoring areas;

[0010] The edge server configures an identification model according to the risk types in each monitoring area, and configures identification resources for the identification tasks corresponding to the risk types according to the weights of the risk types in the monitoring area, so as to identify various discharge hazards. Among them, the risk types with higher weights are allocated more identification resources than those with lower weights.

[0011] In some embodiments, through statistical analysis of the triple dataset, the environmental attribute combinations strongly associated with the risk types are determined. Specifically:

[0012] The environmental attribute combinations involved in the same risk type in the triple dataset are statistically analyzed, and several environmental attributes with an occurrence ratio greater than the threshold are determined to form the environmental attribute combinations strongly associated with the risk type.

[0013] In some embodiments, based on the environmental attributes of the monitoring area, the risk types associated with the monitoring area are determined according to the environmental attribute combinations strongly associated with the risk types, and the weights of the risk types are determined. Specifically:

[0014] The monitoring area is sliced non - overlapping by a rectangular sliding window of a certain size to obtain a plurality of slice units;

[0015] The environmental attributes of the slice units are determined according to the positions of the slice units, and the risk types adapted to the environmental attributes are judged;

[0016] According to the risk types adapted to each slice unit, all the risk types of the monitoring area and the corresponding risk type coverage rates are determined. According to the coverage rate of the risk type and the average severity level corresponding to the risk type, the weight of the risk type is determined;

[0017] Among them, the weight of the risk type where \(i,m\in[1,n]\), \(n\) represents the total number of risk types corresponding to the monitoring area, \(g\) i represents the coverage rate of risk type \(i\), \(h\) i represents the average severity level corresponding to risk type \(i\).

[0018] In some embodiments, the step of allocating each monitoring area to the edge server specifically includes:

[0019] Construct a risk type vector for each monitoring area. The dimension of the risk type vector is set according to the total number of all risk types; for the risk types included in the monitoring area, set it to 1, otherwise set it to 0. The monitoring areas with the same risk type vectors are classified to obtain several sub - classes. Calculate the norm of the risk type vectors of each sub - class as the vector value, and judge whether each sub - class belongs to another sub - class with a larger vector value than itself. If so, mark the subordinate relationship between the two sub - classes;

[0020] Among them, if all the risk types of the subclass with a smaller vector value belong to the risk types of the subclass with a larger vector value, the two subclasses have a subordinate relationship; the edge servers are allocated to each subclass in the order of the vector value size; when it is necessary to allocate the monitoring areas belonging to different subclasses to the same edge server, the subordinate subclasses of the subclass already allocated to the edge server are preferentially allocated.

[0021] In some embodiments, the monitoring area is sliced without overlap according to a rectangular sliding window of a certain size to obtain a plurality of slice units. Specifically:

[0022] Select the size of the slice unit according to the size of the monitoring area;

[0023] Map the monitoring area into the coordinate system, and map the segmentation matrix into the coordinate system according to the size of the slice unit, so that the segmentation matrix overlaps with the monitoring area;

[0024] Determine the units surrounded by the monitoring area in the segmentation matrix as slice units.

[0025] In some embodiments, the following steps are further included:

[0026] According to the request of the user on the browser or the client, load the map, and mark the risk information of each area on the map according to the model prediction results reported by the edge server.

[0027] In some embodiments, the cause of the hidden danger is the result of text description. The semantic features of the cause of the hidden danger are extracted, and the feature vectors after the semantic feature extraction are clustered. Specifically, it includes:

[0028] Extract the semantic features of the cause of the hidden danger through a trained general semantic model and map them into N-dimensional feature vectors;

[0029] Extract the feature vectors of multiple causes of hidden dangers as the initial centers, traverse each sample of the centers and recalculate the center positions, and iterate the required number of times to obtain the final clustering result.

[0030] In some embodiments, the model deployed on the edge server is uniformly trained by the cloud and then sent to be deployed in the edge server.

[0031] On the other hand, the embodiment of the present application provides a cloud-edge collaborative distribution network cable discharge hidden danger identification system, including:

[0032] A memory for storing programs;

[0033] A processor for loading the program to execute the cloud-edge collaborative distribution network cable discharge hidden danger identification method.

[0034] On the other hand, an embodiment of the present application provides a cloud-edge collaborative distribution network cable discharge hidden danger identification system, including a cloud and an edge server;

[0035] The cloud is used to: obtain the severity level, cause and location of the hidden danger event; extract semantic features from the cause of the hidden danger, cluster the feature vectors after semantic feature extraction, perform sampling analysis on the clustering results to obtain keywords, and screen the clustering results based on the keywords, and classify the hidden danger events into multiple risk types; based on the map information, determine the environmental attributes of the occurrence location, so as to obtain a triple data set composed of severity level - risk type - environmental attributes; through statistical analysis of the triple data set, determine the environmental attribute combinations strongly associated with the risk type; based on the environmental attributes of the monitoring area, determine the risk types associated with the monitoring area according to the environmental attribute combinations strongly associated with the risk type, and determine the weights of the risk types; allocate each monitoring area to the edge server, and one edge server serves multiple monitoring areas;

[0036] The edge server is used to configure an identification model according to the risk types in each monitoring area, and configure identification resources for the identification tasks corresponding to the risk types according to the weights of the risk types in the monitoring area, so as to identify various hidden dangers, where the risk type with a higher weight is allocated more identification resources relative to the risk type with a lower weight.

[0037] Beneficial effects: Through this solution, semantic analysis is performed on the severity level, cause and location of the hidden danger events in the historical data, so as to classify the data into multiple risk types that can be identified by the model. Then, based on the map information, the environmental attributes of the occurrence location are determined, so as to obtain a triple data set composed of severity level - risk type - environmental attributes. Subsequently, the weights of various risk types in the monitoring area are analyzed based on the triple. Then, the edge server configures an identification model according to the risk types in each monitoring area, and configures identification resources for the identification tasks corresponding to the risk types according to the weights of the risk types in the monitoring area, so as to identify various discharge hidden dangers, where the risk type with a higher weight is allocated more identification resources relative to the risk type with a lower weight. Through this solution, resources can be optimized according to local conditions, and more resources can be allocated to the types with higher risks, so that they can be monitored more frequently. This solution optimizes the identification efficiency of cable hidden danger discharge in the operation process of the distribution network. Description of the Drawings

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings used in the description of the embodiments.

[0039] Figure 1 It is a flowchart of a method provided by an embodiment of the present application;

[0040] Figure 2 This is a system block diagram provided by an embodiment of the present application. Detailed implementation manners

[0041] To make the objectives, technical solutions, and advantages of the present application clearer, the following will, with reference to the accompanying drawings in the embodiments of the present application, clearly and completely describe the technical solutions of the present application through implementation manners. Apparently, the described embodiments are some, rather than all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0042] Refer to Figure 1 and Figure 2 , an embodiment of the present application provides a cloud-edge collaboration-based hidden danger identification method. As Figure 2 shown, in the system structure of the present application, it includes a cloud and edge servers. The cloud and edge servers can communicate through the Internet of Things or the Internet. The cloud can configure relevant policies for the edge servers. It can be seen from the figure that one edge server can monitor multiple regions, and for different regions, monitoring of different tasks or the same task can be performed. For different regions, different sizes of resources can be allocated for different tasks. Among them, the size of the task in the figure represents the allocated resources. It can be understood that for tasks with higher weights, more resources can be configured, and these resources support the tasks to use higher-performance solutions or execute at higher frequencies.

[0043] It includes the following steps:

[0044] S1. Obtain the severity level, cause, and occurrence location of the hidden danger event. It can be understood that through the analysis of historical reports, the severity level of each event and the description of the cause of the hidden danger are obtained. The cause description is equivalent to a text description in the nature of a briefing, and the occurrence location can be described by map positioning. Usually, when investigating such information, corresponding reports will be formed, and the characteristics of the hidden danger can be obtained through the analysis of historical reports. It can be understood that there are various situations for hidden danger events, but it may not be possible to directly classify the data by type in historical data.

[0045] S2. Extract semantic features from the cause of the hidden danger, cluster the feature vectors after semantic feature extraction, perform sampling analysis on the clustering results to obtain keywords, and screen the clustering results based on the keywords to classify the hidden danger events into multiple risk types.

[0046] It is understandable that since the potential hazard causes are described in a non-standard text-based manner, data structure processing is required. Generally, the sentence patterns and styles of the report descriptions are basically the same, and the content will include the specific causes of events or potential hazard causes. These causes are classified into certain risk types. Therefore, the text is extracted into vector representations in the semantic space through a general semantic model. It is understandable that these semantic models can be BERT models, GPT models, paragraph2vec, doc2vec, etc. In this step, the text is mapped into vector representations in the model semantic space through an NLP model. Accordingly, clustering can be performed on the content based on the vector representations.

[0047] Then, after clustering, the actual keywords involved in the clustering are determined through sampling analysis. This step can find the words used to express the relevant risk types in the causes. After keyword screening, the samples corresponding to the causes can be filtered out, which can prevent the clustering algorithm from classifying different risk types with similar expressions into one category. In this process, since the keywords used in the accident causes of events belonging to a certain risk type cannot be determined in advance, a certain amount of sampling is required after clustering to determine the keywords. It is understandable that classifying the samples corresponding to the risk types helps to further analyze the environmental characteristics of the risk types. It is understandable that the risk type corresponds to at least one type of risk prediction model. It is understandable that the risk prediction models described in this application are existing models, and this application does not improve these models, but only discusses how to allocate resources for these diverse models. These models can be models for predicting the probability of potential hazard discharge caused by tree faults, or models for predicting the degree of potential hazard discharge caused by rain. This application attempts to analyze the relationship between the environmental characteristics of the region and the types of potential hazard discharges, and thus allocate resources corresponding to different risk types accordingly.

[0048] In step S2, semantic feature extraction is performed on the potential hazard causes, and clustering is performed on the feature vectors after semantic feature extraction, specifically including:

[0049] S21. Extract the semantic features of the potential hazard causes through a trained general semantic model and map them into N-dimensional feature vectors. It is understandable that through an NLP model trained in Chinese, the input sentences and paragraphs can be mapped into the semantic space and a semantic vector can be output. It is understandable that the representation based on the vector by the model is actually the weighted superposition of the meanings and order relationships of each word segment. Therefore, the vector output by the model represents the overall meaning, and the length of this vector can output the same length of vector result regardless of the length difference of the input.

[0050] S22. Extract the feature vectors of multiple potential hazard causes as the initial centers, traverse each sample's center and recalculate the center position, iterate the required number of times, and obtain the final clustering result. This solution can adopt the kmean clustering algorithm. First, initialize K centers, then cluster the vectors by calculating the distances between vectors, and update the center position. Then perform iterative optimization until the number of iterations is satisfied.

[0051] S3. Based on the map information, determine the environmental attributes of the occurrence location, so as to obtain a triple dataset composed of severity level - risk type - environmental attributes; through statistical analysis of the triple dataset, determine the environmental attribute combinations strongly associated with the risk type.

[0052] Specifically, through statistical analysis of the triple dataset, determine the environmental attribute combinations strongly associated with the risk type:

[0053] S31. Statistically analyze the environmental attribute combinations involved in the same risk type in the triple dataset, and determine that several environmental attributes with an occurrence proportion greater than the threshold constitute the environmental attribute combinations strongly associated with the risk type.

[0054] It can be understood that by statistically analyzing the environmental attributes with an occurrence proportion of approximately a certain proportion of the same risk type, that is, in the total sample, the occurrence proportion of environmental attributes. One solution is to directly determine all the mutual combinations of these environmental attributes as the environmental attribute combinations. Another solution is to perform secondary screening according to the combination occurrence proportion. After combining the environmental attributes, select those with a combination occurrence proportion greater than the threshold as the strongly associated environmental attribute combinations. It can be understood that strong association means that there is a close relationship between the two. That is, the occurrence of one party has a relatively large relationship with the other party. It can be understood that this step actually determines the environmental attribute combinations according to a certain algorithm.

[0055] S4. Based on the environmental attributes of the monitoring area, determine the risk types associated with the monitoring area according to the environmental attribute combinations strongly associated with the risk type, and determine the weights of the risk types; it can be understood that the environmental attributes include building type, road type, river type, line, greening type, climate type, etc.

[0056] Step S4 specifically includes:

[0057] S41. Slice the monitoring area without overlap using a rectangular sliding window of a certain size to obtain multiple slice units.

[0058] Specifically, select the size of the slicing unit according to the size of the monitoring area; map the monitoring area into a coordinate system, and map the segmentation matrix into the coordinate system according to the size of the slicing unit so that the segmentation matrix overlaps with the monitoring area; determine the units surrounded by the monitoring area in the segmentation matrix as slicing units.

[0059] It can be understood that for a relatively large monitoring area, the environmental elements it contains may be diverse. However, for a small area, this may not be the case. And the environmental elements required for risks need to be concentrated in small areas. Therefore, in this solution, the large area is sliced, and the actual environmental conditions of the slicing units are analyzed. For the monitoring area, it is not defined as a regular area. When analyzing, for the convenience of slicing, it is sliced through a regular shape. Therefore, the monitoring area can be mapped into a coordinate system and then sliced in the way of a rectangular sliding window.

[0060] S42. Determine the environmental attributes of the slicing unit according to the position where the slicing unit is located, and judge the risk types adapted to the environmental attributes. It can be understood that the environmental attributes of each slicing unit can be analyzed, and then the risk types that can exist in the environmental attributes can be matched.

[0061] S43. According to the risk types adapted to each slicing unit, determine all the risk types of the monitoring area and the corresponding risk type coverage rates, and determine the weights of the risk types according to the coverage rates of the risk types and the average severity levels corresponding to the risk types;

[0062] Among them, the weight of the risk type Among them, i and m belong to positive integers in [1, n], n represents the total number of risk types corresponding to the monitoring area, g i represents the coverage rate of risk type i, and h i represents the average severity level corresponding to risk type i.

[0063] It can be understood that for several environmental attributes, as long as the environmental attribute combinations corresponding to the risk types are satisfied, it can be considered that the corresponding risk types may exist in this environment. Similarly, the severity levels corresponding to the same risk type in history can be calculated as described above, so as to calculate the weights of the risk types through the above weight calculation method.

[0064] S5. Allocate each monitoring area to an edge server, and one edge server serves multiple monitoring areas. It can be understood that the monitoring areas are allocated to multiple edge servers to perform regular predictions of the corresponding risk models.

[0065] Specifically, construct a risk type vector for each monitoring area. The dimension of the risk type vector is set according to the total number of all risk types. For the risk types included in the monitoring area, set it to 1, otherwise set it to 0. Classify the monitoring areas with the same risk type vector to obtain several subclasses. Calculate the norm of the risk type vector of each subclass as the vector value. Determine whether each subclass belongs to another subclass with a larger vector value than itself. If so, mark the subordinate relationship between the two subclasses.

[0066] Among them, if all the risk types of the subclass with a smaller vector value belong to the risk types of the subclass with a larger vector value, the two subclasses have a subordinate relationship. Allocate edge servers to each subclass in the order of vector value size. When it is necessary to allocate monitoring areas belonging to different subclasses to the same edge server, give priority to allocating the subordinate subclasses of the subclass already allocated to this edge server.

[0067] Based on the classification results, allocate monitoring tasks to the edge servers in the edge server group. It can be understood that an edge server only needs to allocate the monitoring of all risk types corresponding to its monitoring area. Therefore, when the risk types of the monitoring areas are the same, the number of models configured on the edge server can be reduced. For example, if the risks include four types: A, B, C, and D, the monitoring areas with only risks A and B can be centrally allocated to one edge server. In this way, this server does not need to identify C and D.

[0068] S6. The edge server configures an identification model according to the risk types in each monitoring area, and configures identification resources for the identification tasks corresponding to the risk types according to the weights of the risk types in the monitoring area. Among them, the risk types with higher weights are allocated more identification resources than those with lower weights. It can be understood that several risk identification models will run for each monitoring area, and resources can be allocated according to the weights of each model corresponding to the area. The higher the weight, the higher the identification frequency or the higher the identification priority. The specific strategy can be adjusted according to the situation.

[0069] S7. According to the request of the user on the browser or client, load the map, and mark the risk information of each area on the map according to the model prediction results reported by the edge server. It can be understood that the user logs in through the client, and can display the model prediction results by area in the form of a visual map, and conduct inspections and troubleshoot potential hazards through the above information.

[0070] In some embodiments, the models deployed on the edge servers are uniformly trained by the cloud and then sent down for deployment in the edge servers.

[0071] A cloud-edge collaborative distribution network cable discharge hidden danger identification system includes:

[0072] A memory for storing programs;

[0073] A processor for loading the program to execute a method for identifying potential cable discharge hazards in a distribution network with cloud-edge collaboration. A system for identifying potential cable discharge hazards in a distribution network with cloud-edge collaboration includes a cloud and edge servers;

[0074] The cloud is used to: obtain the severity level, cause of the hazard, and location of occurrence of an accident; extract semantic features from the cause of the hazard, cluster the feature vectors after semantic feature extraction, perform sampling analysis on the clustering results to obtain keywords, and screen the clustering results based on the keywords to classify the hazard events into multiple risk types; based on map information, determine the environmental attributes of the location of occurrence, so as to obtain a triple dataset composed of severity level - risk type - environmental attributes; through statistical analysis of the triple dataset, determine the combination of environmental attributes strongly associated with the risk type; based on the environmental attributes of the monitoring area, determine the risk types associated with the monitoring area according to the combination of environmental attributes strongly associated with the risk type, and determine the weights of the risk types; allocate each monitoring area to an edge server, and one edge server serves multiple monitoring areas;

[0075] The edge server is used to configure an identification model according to the risk types in each monitoring area, and configure identification resources for the identification tasks corresponding to the risk types according to the weights of the risk types in the monitoring area, where the risk types with higher weights are allocated more identification resources compared to those with lower weights.

[0076] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0077] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data that have been authorized by the user or fully authorized by all parties.

[0078] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0079] Note that the above is only the preferred embodiment of the present application and the applied technical principles. Those skilled in the art will understand that the present application is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the present application. Therefore, although the present application has been described in detail through the above embodiments, the present application is not limited to the above embodiments. Without departing from the concept of the present application, it may also include more other equivalent embodiments, and the scope of the present application is determined by the scope of the appended claims.

Claims

1. A method for identifying potential cable discharge hazards in a distribution network with cloud-edge collaboration, characterized in that, Including the following steps: Obtain the severity level, cause of potential hazard, and occurrence location of potential hazard events; Classify potential hazard events into multiple risk types according to the cause of potential hazard; Based on map information, determine the environmental attributes of the occurrence location, so as to obtain a triple dataset composed of severity level - risk type - environmental attributes; through statistical analysis of the triple dataset, determine the environmental attribute combinations strongly associated with the risk type; Based on the environmental attributes of the monitoring area, determine the risk types associated with the monitoring area according to the environmental attribute combinations strongly associated with the risk type, and determine the weights of the risk types; Allocate each monitoring area to an edge server, and one edge server serves multiple monitoring areas; The edge server configures an identification model according to the risk types in each monitoring area, and configures identification resources for the identification tasks corresponding to the risk types according to the weights of the risk types in the monitoring area, so as to identify various types of discharge potential hazards, where the risk type with a higher weight is allocated more identification resources compared to the one with a lower weight; The step of allocating each monitoring area to an edge server specifically includes: Construct a risk type vector for each monitoring area, and the dimension of the risk type vector is set according to the total number of all risk types; set 1 for the risk types included in the monitoring area, otherwise set 0, classify the monitoring areas with the same risk type vector, obtain several subclasses, calculate the norm of the risk type vector of each subclass as the vector value, and judge whether each subclass belongs to another subclass with a larger vector value than itself. If so, mark the subordination relationship between the two subclasses; Among them, if all the risk types of the subclass with a smaller vector value belong to the risk types of the subclass with a larger vector value, the two subclasses have a subordination relationship; allocate the subclasses to the edge servers in the order of the vector values; when it is necessary to allocate monitoring areas belonging to different subclasses to the same edge server, give priority to allocating the subordinate subclasses of the subclass already allocated to the edge server.

2. The method according to claim 1, wherein Through statistical analysis of the triple dataset, determine the environmental attribute combinations strongly associated with the risk type, specifically: Statistically analyze the environmental attribute combinations involved in the triple dataset with the same risk type, and determine that several environmental attributes with an occurrence proportion greater than the threshold constitute the environmental attribute combinations strongly associated with the risk type.

3. The method according to claim 1, characterized in that Based on the environmental attributes of the monitoring area, determine the risk types associated with the monitoring area according to the environmental attribute combinations strongly associated with the risk type, and determine the weights of the risk types, specifically: Slice the monitoring area without overlap according to a rectangular sliding window of a certain size to obtain multiple slice units; Determine the environmental attributes of the slice unit according to the location where the slice unit is located, and judge the risk type suitable for the environmental attributes; According to the risk types suitable for each slice unit, determine all the risk types of the monitoring area and the corresponding risk type coverage rate, and determine the weights of the risk types according to the coverage rate of the risk type and the average severity level corresponding to the risk type; Among them, the weight of the risk type Among them, i and m are positive integers belonging to [1, n], where n represents the total number of risk types corresponding to the monitoring area, and g i represents the coverage rate of risk type i, and h i represents the average severity level corresponding to risk type i.

4. The method according to claim 3, characterized in that, Slice the monitoring area without overlap according to a rectangular sliding window of a certain size to obtain multiple slice units: Select the size of the slice unit according to the size of the monitoring area; Map the monitoring area into a coordinate system, and map the segmentation matrix into the coordinate system according to the size of the slice unit, so that the segmentation matrix overlaps with the monitoring area; Determine the units surrounded by the monitoring area in the segmentation matrix as slice units.

5. The method according to claim 1, characterized in that, According to the hidden danger causes, classify the hidden danger events into multiple risk types, specifically including: Extract the semantic features of the hidden danger causes, cluster the feature vectors after semantic feature extraction, conduct sampling analysis on the clustering results to obtain keywords, and based on the keywords, screen the clustering results, and classify the hidden danger events into multiple risk types.

6. The method according to claim 5, wherein The hidden danger cause is the result of text description. The extraction of semantic features of the hidden danger cause and the clustering of the feature vectors after semantic feature extraction specifically include: Extract the semantic features of the hidden danger cause through a trained general semantic model and map them into an N-dimensional feature vector; Extract the feature vectors of multiple hidden danger causes as the initial centers, traverse the centers of each sample and recalculate the center positions, and iterate the required number of times to obtain the final clustering result.

7. The method according to any one of claims 1-5, characterized in that The model deployed on the edge server is uniformly trained by the cloud and then sent to be deployed in the edge server.

8. A cloud-edge collaborative hidden danger identification system for cable discharge in a distribution network, characterized in that, Including: A memory for storing programs; A processor for loading the program to execute the method for identifying hidden dangers of cable discharge in a cloud-edge collaborative distribution network as described in any one of claims 1-7.

9. A cloud-edge collaborative hidden danger identification system for cable discharges in a distribution network, characterized in that, Including the cloud and edge servers; The cloud is used for: obtaining the severity level, hidden danger cause and occurrence location of the hidden danger event; classifying the hidden danger event into multiple risk types according to the hidden danger cause; Based on the map information, determine the environmental attributes of the occurrence location, so as to obtain a triple dataset composed of severity level - risk type - environmental attributes; through statistical analysis of the triple dataset, determine the environmental attribute combinations strongly associated with the risk type; based on the environmental attributes of the monitoring area, determine the risk types associated with the monitoring area according to the environmental attribute combinations strongly associated with the risk type, and determine the weights of the risk types; allocate each monitoring area to the edge server, and one edge server serves multiple monitoring areas; The edge server is used to configure an identification model according to the risk types in each monitoring area, and configure identification resources for the identification tasks corresponding to the risk types according to the weights of the risk types in the monitoring area, so as to identify various hidden dangers, where the risk type with a higher weight is allocated more identification resources than the risk type with a lower weight; The allocation of each monitoring area to the edge server specifically includes: Construct a risk type vector for each monitoring area. The risk type vector is set with dimensions according to the total number of all risk types; set 1 for the risk types included in the monitoring area, otherwise set 0. Classify the monitoring areas with the same risk type vector to obtain several subclasses. Calculate the norm of the risk type vector of each subclass as the vector value. Determine whether each subclass belongs to another subclass with a larger vector value than itself. If so, mark the subordination relationship between the two subclasses; Among them, if all the risk types of the subclass with a smaller vector value belong to the risk types of the subclass with a larger vector value, the two subclasses have a subordinate relationship; the edge servers are allocated to each subclass in the order of the vector value size; when it is necessary to allocate monitoring areas belonging to different subclasses to the same edge server, the subordinate subclasses of the subclass already allocated to this edge server are preferentially allocated.

Images