An extreme weather power distribution network fault risk assessment method and system
By collecting and quantifying distribution network data under extreme weather conditions, and combining classification algorithms and fuzzy vectors for risk assessment, the problem of insufficient assessment efficiency and accuracy in existing technologies has been solved, enabling efficient and accurate identification and early warning of distribution network fault risks under extreme weather conditions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FUJIAN ELECTRIC POWER CO LTD XIAMEN ELECTRIC POWER SUPPLY CO
- Filing Date
- 2026-03-12
- Publication Date
- 2026-06-19
Smart Images
Figure CN122246700A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of power risk assessment technology, and specifically relates to a method and system for assessing the risk of power distribution network failures during extreme weather. Background Technology
[0002] Extreme weather has become a major factor affecting the safe and stable operation of power distribution networks. Disasters such as rainstorms, typhoons, blizzards, freezing rain, and lightning can easily cause line tripping, equipment insulation failure, and tower damage, seriously threatening power supply reliability. Existing methods for assessing the risk of power distribution network faults under extreme weather conditions typically collect and analyze single types of data, such as meteorological data, equipment operation data, fault history data, grid environment data, and load data. Moreover, most assessment methods do not quantify and filter the amount of distribution information data under extreme weather conditions, making them susceptible to redundant and interfering data, thus limiting the efficiency and accuracy of the assessment. At the same time, existing methods rarely incorporate the inherent characteristics of the power distribution network to build differentiated assessment models. They lack specificity in identifying fault risks for power distribution networks with different grid structures and load characteristics, making it difficult to accurately depict the correlation between extreme weather and fault occurrence. The practicality and reliability of the risk assessment results need to be improved. Summary of the Invention
[0003] To address the problems existing in the prior art, this invention proposes a method and system for assessing the risk of power distribution network failures during extreme weather.
[0004] The technical solution of the present invention is as follows: On the one hand, the present invention provides a method for assessing the risk of power distribution network failures during extreme weather, comprising the following steps: Obtain a dataset of potential faults for each distribution network generated during the operation of the distribution network, including meteorological data, distribution network equipment operation data, historical distribution network fault data, distribution network structure and environmental data, and load-related data; The amount of extreme weather power distribution information data obtained during the operation of each power distribution network was obtained by filtering and quantifying the dataset of potential faults in the power distribution network. Based on the amount of extreme weather power distribution information data, a target fault and hazard dataset is selected from all fault and hazard datasets; The target fault hazard dataset is processed by the distribution network risk type assessment thread to obtain the risk assessment results.
[0005] Preferably, the meteorological data includes: extreme weather and regular weather data, with extreme weather data covering key indicators such as rainfall and duration of heavy rain, wind speed and wind pressure of typhoons, snowfall and snow thickness of blizzards, ice thickness and duration of low temperatures of freezing, and lightning strike density and lightning current amplitude. Regular weather data includes auxiliary indicators such as daily average temperature, humidity, and wind direction. Distribution network equipment operation data includes: real-time operating parameters, insulation performance data, mechanical performance data, load carrying capacity data, and basic parameters of various types of equipment. Distribution network fault history data includes: fault occurrence time, fault type, fault location, fault cause, fault duration, repair records, and fault recurrence status of various types of equipment. Distribution network structure and environmental data include: network topology, line routing, tower distribution, equipment installation environment parameters, and grounding resistance data. Load-related data includes: real-time distribution of various loads, peak / valley load data, power consumption characteristics of loads in important areas, and load priority data.
[0006] Preferably, the process of selecting the target fault hazard dataset from all fault hazard datasets based on the amount of extreme weather power distribution information data specifically involves: For the fault hazard datasets containing specified redundant interference features, delete the specified redundant interference features to obtain optimized fault hazard datasets. The optimized datasets of potential faults are compared pairwise, and the datasets in which the difference between the Chinese text in the pairwise comparisons is greater than a preset threshold are selected as the datasets of potential faults. Based on the amount of extreme weather power distribution information data corresponding to the dataset of undetermined potential faults, the dataset with the largest amount of extreme weather power distribution information data is selected from the dataset of undetermined potential faults as the target dataset of potential faults.
[0007] Preferably, the training method for the distribution network risk type assessment thread is as follows: Obtain a dataset of training fault hazards for each distribution network generated during the operation of the training event, as well as the preset distribution network types for the distribution network in operation of the training event; Feature extraction processing is performed on the training fault hazard dataset to obtain key features; For each preset distribution network type, feature extraction processing is performed to obtain the type feature vector of each preset distribution network type; The distribution network risk type assessment thread is obtained by training based on the key features, the type feature vector of each preset distribution network type, and the preset distribution network type.
[0008] Preferably, the step of training based on the key features, the category feature vector of each preset distribution network type, and the preset distribution network type to obtain the distribution network risk type assessment thread specifically involves: The key features are classified using a classification algorithm to obtain the risk type regression analysis results of the power distribution network operation for the training items; For each preset distribution network type, the type feature vector, the key features, and the preset distribution network type are classified using a classification method based on feature similarity and fuzzy vectors to obtain the risk type simulation results of the distribution network for the training item. The difference between the simulation results and the regression analysis results of the risk type is calculated using a difference metric function, and the training is performed in the direction of reducing the difference in the probability distribution. The process is terminated and output when the difference metric function value is less than a preset threshold or the number of training iterations reaches a preset maximum value, thus obtaining the distribution network risk type assessment thread.
[0009] Preferably, the risk type simulation results of the power distribution network for each preset power distribution network type are obtained by using a classification method based on feature similarity and fuzzy vectors for the type feature vector, the key features, and the preset power distribution network type, as follows: The sharing coefficient between the category feature vector of each preset distribution network type and the key feature is calculated based on the vector similarity function, and a catalog fuzzy vector is generated based on the obtained sharing coefficients using the Gaussian membership function. The directory fuzzy vector is compressed to obtain the directory feature vector; The directory fuzzy vector and the directory feature vector are concatenated, and the concatenated fused vector is classified using a classification algorithm to obtain the simulation results of the risk types of the power distribution network for the training items.
[0010] On the other hand, the present invention provides an extreme weather power distribution network fault risk assessment system, including a data acquisition module, a data volume acquisition module, a dataset filtering module, and a risk assessment module; The data acquisition module is used to obtain the dataset of potential faults in each distribution network generated during the operation of the distribution network. This dataset includes meteorological data, distribution network equipment operation data, historical distribution network fault data, distribution network structure and environmental data, and load-related data. The data acquisition module is used to filter and quantify the distribution network fault hazard dataset to obtain the amount of extreme weather power distribution information data obtained during the operation of each distribution network. The dataset filtering module is used to filter the target fault hazard dataset from all fault hazard datasets based on the amount of extreme weather power distribution information data; The risk assessment module is used to perform risk type assessment on the target fault hazard dataset through the distribution network risk type assessment thread to obtain risk assessment results.
[0011] In another aspect, the present invention also provides an electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method as described in any embodiment of the present invention.
[0012] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any embodiment of the present invention.
[0013] Compared with the prior art, the present invention has the following technical effects: This invention first uses a classification algorithm to learn and reason about key features to obtain a probability distribution of risk types that conforms to historical operating patterns. Then, based on feature similarity and fuzzy vectors, it performs fuzzy reasoning on different preset distribution network types, type feature vectors, and key features to form a simulated distribution of risk types that fits the power grid topology and disaster mechanism. By quantifying and iteratively reducing the difference in probability distribution between the two types of distributions through a difference measurement function, it finally outputs a stable and convergent risk type assessment thread, which can accurately distinguish the differentiated impact of different disaster types such as typhoons, rainstorms, and icing on lines, towers, and distribution equipment. This significantly improves the accuracy, robustness, and generalization ability of distribution network risk type identification, and provides an efficient and reliable intelligent assessment method for the quantitative assessment and early warning of distribution network operation risks in multi-hazard scenarios. Attached Figure Description
[0014] Figure 1 This is an overall flowchart of the extreme weather power distribution network fault risk assessment method described in this invention. Detailed Implementation
[0015] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments of the present application and with reference to the accompanying drawings.
[0016] Example 1 This embodiment provides a method for assessing the risk of power distribution network failures during extreme weather. (See also...) Figure 1 As shown, it includes the following steps: The dataset of potential faults in each distribution network generated during the operation of the distribution network includes meteorological data, distribution network equipment operation data, historical distribution network fault data, distribution network structure and environmental data, and load-related data.
[0017] In a preferred embodiment of this invention, the meteorological data includes: extreme weather and regular weather data. Extreme weather data covers key indicators such as rainfall and duration of heavy rain, wind speed and pressure of typhoons, snowfall and snow depth of blizzards, icing thickness and duration of low temperatures of freezing events, and lightning strike density and lightning current amplitude. Regular weather data includes auxiliary indicators such as daily average temperature, humidity, and wind direction. The power distribution network equipment operation data includes: real-time operating parameters, insulation performance data, mechanical performance data, load-bearing data, and other parameters of various types of equipment (such as overhead lines, towers, transformers, switches, etc.). It has its own basic parameters (such as model, years of operation, conductor cross-section, etc.); historical data of distribution network faults include: the time of occurrence of various equipment faults, fault type, fault location, fault cause, fault duration, emergency repair records and fault recurrence; data on distribution network structure and environment include: network topology, line route, tower distribution, equipment installation environment parameters (such as terrain, surrounding obstacles), grounding resistance data, etc.; load-related data include: real-time distribution of various loads, peak / valley load data, power consumption characteristics and load priority data of important area loads (such as hospitals, transportation hubs, etc.).
[0018] The amount of extreme weather power distribution information data obtained during the operation of each power distribution network was obtained by filtering and quantifying the dataset of potential faults in the power distribution network.
[0019] Specifically, the data set for screening potential faults in the power distribution network is as follows: Data extraction: Extract two types of related data from the fault and potential hazard dataset, such as core data on extreme weather (rainfall / duration of heavy rain, wind speed / pressure of typhoons, snowfall / snow thickness of blizzards, ice thickness / duration of low temperature during freezing, and lightning strike density / lightning current amplitude); or power distribution related data under extreme weather conditions (operating parameters of power distribution equipment, fault records, load data, grid operation data, etc.).
[0020] Determine the statistical scope: Use a single valid data record as the statistical unit (e.g., 1 hourly rainfall data + corresponding time period line load data is counted as 1 valid related data), and remove invalid and redundant data (e.g., extreme weather records that are missing corresponding power distribution data).
[0021] Quantitative calculation: The total number of valid data entries is the data volume. For precise storage, the scale can be multiplied by the number of bytes per data entry.
[0022] Based on the amount of extreme weather power distribution information data, a target fault and hazard dataset is selected from all fault and hazard datasets.
[0023] As a preferred embodiment of this practice, the specific method for selecting the target fault hazard dataset from all fault hazard datasets based on the extreme weather power distribution information data volume is as follows: For the fault hazard datasets containing specified redundant interference features (redundant / invalid / interference elements that need to be deleted without affecting fault risk assessment, such as irrelevant equipment logs, miscellaneous environmental data not related to power distribution, etc.), the specified redundant interference features are deleted to obtain several optimized fault hazard datasets.
[0024] The optimized datasets of potential faults are compared pairwise, and the datasets with text differences greater than a preset threshold are selected as potential fault datasets.
[0025] Based on the amount of extreme weather power distribution information data corresponding to the pending fault hazard dataset, the dataset with the largest amount of extreme weather power distribution information data is selected as the target fault hazard dataset. The larger the amount of extreme weather power distribution information data obtained during the operation of the power distribution network, the higher the importance of the power distribution network transaction to the power distribution network matters. Correspondingly, the fault hazard dataset of the power distribution network can more directly reflect the actual operation activities of the power distribution network matters.
[0026] The target fault hazard dataset is processed by the distribution network risk type assessment thread to obtain the risk assessment results.
[0027] As a preferred embodiment of this practice, the training method for the distribution network risk type assessment thread is as follows: The system obtains a dataset of training fault hazards for each distribution network generated during the operation of the training event, as well as the preset distribution network types for the training event's operating distribution network. Specifically, the preset distribution network types may include: by grid structure: overhead distribution network, cable distribution network, hybrid distribution network, etc.; by voltage level: medium-voltage distribution network (10kV), low-voltage distribution network (0.4kV), etc.; by regional environment: urban distribution network, rural distribution network, mountainous distribution network (adapted to different extreme weather impact scenarios), etc.; by load characteristics: industrial load distribution network, residential load distribution network, comprehensive load distribution network (including important loads such as hospitals and transportation hubs), etc.; by equipment configuration: conventional distribution network, smart distribution network (including distribution automation and online monitoring equipment), etc.
[0028] Feature extraction is performed on the training fault hazard dataset to obtain key features. The specific operations of the feature extraction process are as follows: Data preprocessing: The training dataset for potential faults is cleaned (invalid and redundant data are removed), normalized, and the data format is standardized to eliminate the influence of units.
[0029] Feature screening: Screen core features related to distribution network faults, eliminate irrelevant features, and focus on retaining extreme weather features (rainfall, wind speed, etc.), equipment operation features, fault features, and load features.
[0030] Feature transformation: Feature engineering methods (such as encoding and dimensionality reduction) are used to transform non-numerical features (such as equipment model and fault type) into trainable numerical features, simplifying feature dimensions.
[0031] Core feature output: Extract the key features after processing to form a standardized feature set, which will be used for subsequent training related to distribution network faults.
[0032] For each preset distribution network type, feature extraction processing is performed to obtain a type feature vector for each preset distribution network type. The specific operation of the feature extraction processing is as follows: for a single preset distribution network type (such as overhead distribution network or urban distribution network), the core features specific to that type are extracted to generate the corresponding type feature vector. It is necessary to extract features specifically based on the characteristics of the type.
[0033] The distribution network risk type assessment thread is obtained by training based on the key features, the type feature vector of each preset distribution network type, and the preset distribution network type.
[0034] In a preferred embodiment of this practice, the step of training based on the key features, the category feature vector of each preset distribution network type, and the preset distribution network type to obtain the distribution network risk type assessment thread specifically involves: The key features are classified using a classification algorithm to obtain the risk category regression analysis results of the power distribution network for the training items. These results include risk category probability distribution vectors for each preset power distribution network type. Each element in the vector corresponds to the probability of belonging to a preset power distribution network risk type, and the sum of all element probabilities is 1. Specifically, the classification algorithm preferentially uses multinomial logistic regression. If the preset power distribution network types have significant differences in features, multi-classification algorithms such as Support Vector Machine (SVM) and Random Forest can also be used. The core objective is to output the probability vectors for each risk category, which includes risks such as line galloping and wind-induced flashover, tower tilting, collapse, and foundation scour, equipment insulation failure, tower icing and overload, line phase-to-phase short circuits and equipment thermal aging, and power distribution network cascading failures, etc.
[0035] The risk type simulation results of the power distribution network for each preset power distribution network type are obtained by using a classification method based on feature similarity and fuzzy vectors for the type feature vector, the key features, and the preset power distribution network type.
[0036] The difference between the simulation results and regression analysis results of the risk types is calculated using a difference metric function. Training is then performed in the direction of reducing this difference. The process terminates and outputs the result when the difference metric function value is less than a preset threshold (i.e., the distribution difference reaches an acceptable range) or the number of training iterations reaches a preset maximum (to prevent overtraining). This results in the distribution network risk type assessment thread. Specifically, the difference metric function preferentially uses KL divergence (relative entropy), which is primarily used to measure the difference between the two probability distributions (adapting the output characteristics of the probability vector). Cross-entropy and JS divergence can also be used, both quantifying the distribution difference between the simulation results and the regression analysis results. The training direction is to make the distribution of the regression analysis results closer to the simulation results, as the simulation results are accurate category matching results obtained based on feature vectors, key features, and preset distribution network types, serving as the training benchmark. No synchronous evaluation is required after training. After training, the distribution network risk type assessment thread integrates the core logic of both methods, and evaluation can be performed solely through this thread. The distribution network risk type assessment thread will directly output a single assessment result, which is essentially a risk type regression analysis result optimized after training. It selects the category with the highest probability of belonging as the final distribution network risk type assessment result, balancing accuracy and efficiency.
[0037] In a preferred embodiment, a classification method based on feature similarity and fuzzy vectors is applied to the category feature vector, the key features, and the preset distribution network type for each preset distribution network type to obtain the specific simulation results of the risk types of the distribution network for the training item: The sharing coefficient between the category feature vector and the key feature of each preset distribution network type is calculated based on the vector similarity function. Then, a directory fuzzy vector is generated based on the obtained sharing coefficients using a Gaussian membership function. Specifically, the sharing coefficient calculation measures the overlap between the category feature vector and the key feature. In this embodiment, a cosine similarity function is preferred, but the Pearson correlation coefficient can also be used. The directory fuzzy vector is calculated as follows: based on all sharing coefficients, a Gaussian membership function (preferred) is used to map them to membership values in the [0,1] interval. These values are then arranged according to the preset distribution network type order to obtain the directory fuzzy vector, which is used for subsequent fuzzy matching evaluation.
[0038] The preset distribution network types are subjected to directory compression processing to obtain directory feature vectors. Further, the specific operations of directory compression processing are as follows: Preprocessing: extracting effective membership values from the directory fuzzy vectors and removing invalid values; Compression: using PCA or LDA dimensionality reduction functions to retain core information and remove redundancy; Output: standardizing the compressed vector to obtain directory feature vectors for evaluating thread optimization training.
[0039] The directory fuzzy vector and the directory feature vector are concatenated, and the concatenated fused vector is classified using a classification algorithm (the classification algorithm can be a multinomial logistic regression SVM algorithm) to obtain the simulation results of the risk types of the power distribution network for the training items.
[0040] Example 2 Accordingly, this embodiment provides an extreme weather power distribution network fault risk assessment system to implement the method described in any embodiment of the present invention, including a data acquisition module, a data volume acquisition module, a dataset filtering module, and a risk assessment module; The data acquisition module is used to obtain the dataset of potential faults in each distribution network generated during the operation of the distribution network. This dataset includes meteorological data, distribution network equipment operation data, historical distribution network fault data, distribution network structure and environmental data, and load-related data. The data acquisition module is used to filter and quantify the distribution network fault hazard dataset to obtain the amount of extreme weather power distribution information data obtained during the operation of each distribution network. The dataset filtering module is used to filter the target fault hazard dataset from all fault hazard datasets based on the amount of extreme weather power distribution information data; The risk assessment module is used to perform risk type assessment on the target fault hazard dataset through the distribution network risk type assessment thread to obtain risk assessment results.
[0041] Example 3 This embodiment provides an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the method described in any embodiment of the present invention.
[0042] Example 4 This embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any embodiment of the present invention.
[0043] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.
[0044] Those skilled in the art will recognize that the units and algorithm steps described in the embodiments disclosed herein can be implemented using electronic hardware, computer software, or a combination of electronic hardware and software. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0045] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0046] In the several embodiments provided in this application, any function, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0047] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.
Claims
1. A method for assessing the risk of power distribution network faults during extreme weather, characterized in that, Includes the following steps: Obtain a dataset of potential faults for each distribution network generated during the operation of the distribution network, including meteorological data, distribution network equipment operation data, historical distribution network fault data, distribution network structure and environmental data, and load-related data; The amount of extreme weather power distribution information data obtained during the operation of each power distribution network was obtained by filtering and quantifying the dataset of potential faults in the power distribution network. Based on the amount of extreme weather power distribution information data, a target fault and hazard dataset is selected from all fault and hazard datasets; The target fault hazard dataset is processed by the distribution network risk type assessment thread to obtain the risk assessment results.
2. The method for assessing the risk of power distribution network failures in extreme weather according to claim 1, characterized in that, The meteorological data includes: extreme weather and regular weather data. Extreme weather data covers key indicators such as rainfall and duration of heavy rain, wind speed and wind pressure of typhoons, snowfall and snow depth of blizzards, ice thickness and duration of low temperatures of freezing, and lightning density and lightning current amplitude. Regular weather data includes auxiliary indicators such as daily average temperature, humidity, and wind direction. Distribution network equipment operation data includes: real-time operating parameters, insulation performance data, mechanical performance data, load carrying capacity data, and basic parameters of various types of equipment. Distribution network fault history data includes: fault occurrence time, fault type, fault location, fault cause, fault duration, repair records, and fault recurrence status of various types of equipment. Distribution network structure and environmental data include: network topology, line routing, tower distribution, equipment installation environment parameters, and grounding resistance data. Load-related data includes: real-time distribution of various loads, peak / valley load data, power consumption characteristics of loads in important areas, and load priority data.
3. The method for assessing the risk of power distribution network failures in extreme weather according to claim 1, characterized in that, The specific steps for selecting the target fault and hazard dataset from all fault and hazard datasets based on the extreme weather power distribution information data volume are as follows: For the fault hazard datasets containing specified redundant interference features, delete the specified redundant interference features to obtain optimized fault hazard datasets. The optimized datasets of potential faults are compared pairwise, and the datasets in which the difference between the Chinese text in the pairwise comparisons is greater than a preset threshold are selected as the datasets of potential faults. Based on the amount of extreme weather power distribution information data corresponding to the dataset of undetermined potential faults, the dataset with the largest amount of extreme weather power distribution information data is selected from the dataset of undetermined potential faults as the target dataset of potential faults.
4. The method for assessing the risk of power distribution network failures in extreme weather according to claim 1, characterized in that, The training method for the distribution network risk type assessment thread is as follows: Obtain a dataset of training fault hazards for each distribution network generated during the operation of the training event, as well as the preset distribution network types for the distribution network in operation of the training event; Feature extraction processing is performed on the training fault hazard dataset to obtain key features; For each preset distribution network type, feature extraction processing is performed to obtain the type feature vector of each preset distribution network type; The distribution network risk type assessment thread is obtained by training based on the key features, the type feature vector of each preset distribution network type, and the preset distribution network type.
5. The method for assessing the risk of power distribution network failures in extreme weather according to claim 4, characterized in that, The process of training based on the key features, the category feature vector of each preset distribution network type, and the preset distribution network type to obtain the distribution network risk type assessment thread is as follows: The key features are classified using a classification algorithm to obtain the risk type regression analysis results of the power distribution network operation for the training items; For each preset distribution network type, the type feature vector, the key features, and the preset distribution network type are classified using a classification method based on feature similarity and fuzzy vectors to obtain the risk type simulation results of the distribution network for the training item. The difference between the simulation results and the regression analysis results of the risk type is calculated using a difference metric function, and the training is performed in the direction of reducing the difference in the probability distribution. The process is terminated and output when the difference metric function value is less than a preset threshold or the number of training iterations reaches a preset maximum value, thus obtaining the distribution network risk type assessment thread.
6. The method for assessing the risk of power distribution network failures in extreme weather according to claim 5, characterized in that, The risk simulation results of the power distribution network operation for each preset power distribution network type are obtained by applying a classification method based on feature similarity and fuzzy vectors to the type feature vector, the key features, and the preset power distribution network type. The sharing coefficient between the category feature vector of each preset distribution network type and the key feature is calculated based on the vector similarity function, and a catalog fuzzy vector is generated based on the obtained sharing coefficients using the Gaussian membership function. The directory fuzzy vector is compressed to obtain the directory feature vector; The directory fuzzy vector and the directory feature vector are concatenated, and the concatenated fused vector is classified using a classification algorithm to obtain the simulation results of the risk types of the power distribution network for the training items.
7. A power distribution network fault risk assessment system for extreme weather, characterized in that, The system is used to implement the method as described in any one of claims 1 to 6, and includes a data acquisition module, a data volume acquisition module, a dataset filtering module, and a risk assessment module; The data acquisition module is used to obtain the dataset of potential faults in each distribution network generated during the operation of the distribution network. This dataset includes meteorological data, distribution network equipment operation data, historical distribution network fault data, distribution network structure and environmental data, and load-related data. The data acquisition module is used to filter and quantify the distribution network fault hazard dataset to obtain the amount of extreme weather power distribution information data obtained during the operation of each distribution network. The dataset filtering module is used to filter the target fault hazard dataset from all fault hazard datasets based on the amount of extreme weather power distribution information data; The risk assessment module is used to perform risk type assessment on the target fault hazard dataset through the distribution network risk type assessment thread to obtain risk assessment results.
8. An electronic device, the electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the method as described in any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.