Power grid data reconstruction method and device, electronic equipment and storage medium

Abstract

The invention discloses a power grid data reconstruction method and device, electronic equipment and a storage medium, and relates to a data processing technology, the method comprises the steps that a sample data set is constructed according to collected power grid accident information, the power grid accident information comprises at least two accident dimensions, and each accident dimension comprises multiple pieces of accident data; performing feature extraction on the accident data corresponding to the at least two accident dimensions to obtain an accident feature corresponding to each accident dimension; training a preset network model based on the accident feature corresponding to each accident dimension to obtain a data reconstruction model; and inputting the accident text corresponding to the target power grid accident into the data reconstruction model to obtain missing data of the target power grid accident. According to the method, the missing data corresponding to any power grid accident can be output through the trained data reconstruction model, so that the integrity of the accident data is ensured, and when the early warning model is trained according to the complete accident data, the accuracy and the reliability of an output result of the early warning model can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a power grid data reconstruction method, device, electronic device and storage medium. Background Art

[0002] In my country's power industry, power grid safety is directly related to the stability of industrial production and people's lives. Traditionally, in order to reduce the incidence of power accidents, power companies usually improve the safety of the power grid by strengthening employee safety education and equipment maintenance. However, due to the high risk of the power industry itself, these traditional methods have not been able to completely solve the problem of frequent power accidents.

[0003] In order to provide safety warnings for power accidents and reduce the accident rate, warning models are usually established by analyzing relevant data of power grid accidents. The warning models can effectively prevent and promptly handle power grid accidents. However, accurate warning models rely on complete accident data. In actual situations, different power grid companies have different dimensional requirements and focuses on accident records. Therefore, when integrating historical power grid accident data for training safety warning models, some power grid data will be missing. As a result, the established warning model has the problem of inaccurate output results due to the missing of some power grid data. Summary of the invention

[0004] The present application provides a power grid data reconstruction method, device, electronic device and storage medium, which can ensure the integrity of power grid accident data and improve the output accuracy of early warning models.

[0005] In a first aspect, the present application provides a power grid data reconstruction method, comprising: constructing a sample data set based on collected power grid accident information, the power grid accident information including at least two accident dimensions, each of the accident dimensions including multiple accident data; performing feature extraction on the accident data corresponding to at least two of the accident dimensions to obtain accident features corresponding to each of the accident dimensions; training a preset network model based on the accident features corresponding to each of the accident dimensions to obtain a data reconstruction model; inputting the accident text corresponding to the target power grid accident into the data reconstruction model to obtain the missing data of the target power grid accident.

[0006] Optionally, constructing a sample data set based on the collected power grid accident information includes: obtaining accident text related to the power grid accident from at least one type of web page address according to a preset corpus; classifying the text content in the accident text based on the accident dimension to obtain accident data corresponding to each of the accident dimensions; and constructing the sample data set according to the accident data corresponding to each of the accident dimensions.

[0007] Optionally, constructing the sample data set according to the accident data corresponding to each of the accident dimensions includes: determining the number of fields of the accident field corresponding to each of the accident dimensions in the accident data; performing enhancement processing on the corresponding sample data according to the number of fields corresponding to each of the accident dimensions to obtain enhanced sample data corresponding to each of the accident dimensions; and constructing the sample data set according to the accident text of each of the accident dimensions and the corresponding enhanced sample data.

[0008] Optionally, the feature extraction of the accident data corresponding to at least two of the accident dimensions to obtain the accident features corresponding to each of the accident dimensions includes: performing sequence processing on each of the accident data to obtain a character sequence corresponding to each of the accident data; performing a convolution operation on the character sequence corresponding to each of the accident data based on a convolutional neural network model to obtain a local feature corresponding to each of the accident data; and determining the accident features of the corresponding accident dimension based on the local feature corresponding to each of the accident data.

[0009] Optionally, the preset network model is trained based on the accident features corresponding to each of the accident dimensions to obtain the data reconstruction model, including: determining the dependency relationship between every two accident dimensions based on the accident features corresponding to each of the accident dimensions; training the preset network model according to the dependency relationship between every two accident dimensions, and obtaining the data reconstruction model when the training results converge.

[0010] Optionally, the step of inputting the accident text corresponding to the target power grid accident into the data reconstruction model to obtain the missing data of the target power grid accident includes: obtaining known accident data corresponding to known accident dimensions from the accident text; determining the accident dimensions to be reconstructed based on the known accident dimensions; inputting the known accident data corresponding to the known accident dimensions and the accident dimensions to be reconstructed into the data reconstruction model to obtain the missing data corresponding to the accident dimensions to be reconstructed.

[0011] In a second aspect, the present application provides a power grid data reconstruction device, the device comprising:

[0012] A data set construction module, used to construct a sample data set according to the collected power grid accident information, wherein the power grid accident information includes at least two accident dimensions, and each of the accident dimensions includes a plurality of accident data;

[0013] A feature extraction module, used to extract features from the accident data corresponding to at least two of the accident dimensions, to obtain accident features corresponding to each of the accident dimensions;

[0014] A model training model, used to train a preset network model based on the accident features corresponding to each of the accident dimensions to obtain a data reconstruction model;

[0015] The data reconstruction module is used to input the accident text corresponding to the target power grid accident into the data reconstruction model to obtain the missing data of the target power grid accident.

[0016] In a third aspect, the present application further provides an electronic device, the electronic device comprising:

[0017] at least one processor; and

[0018] a memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the power grid data reconstruction method described in any embodiment of the present application.

[0020] In a fourth aspect, the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the power grid data reconstruction method described in any embodiment of the present application when executed.

[0021] In a fifth aspect, the present application further provides a computer program product, including a computer program, which, when executed by a processor, implements the power grid data reconstruction method described in any embodiment of the present application.

[0022] The power grid data reconstruction scheme provided in the embodiment of the present application first constructs a sample data set based on the collected power grid accident information, where the power grid accident information includes at least two accident dimensions, and each accident dimension includes multiple accident data; then, feature extraction is performed on the accident data corresponding to at least two accident dimensions to obtain the accident features corresponding to each accident dimension; then, the preset network model is trained based on the accident features corresponding to each accident dimension to obtain a data reconstruction model; finally, the accident text corresponding to the target power grid accident is input into the data reconstruction model to obtain the missing data of the target power grid accident. The data reconstruction model constructed by this scheme can learn the accident data corresponding to multiple accident dimensions, so that the data reconstruction model completed through training can output the missing data corresponding to any power grid accident to ensure the integrity of the accident data, thereby helping to improve the accuracy and reliability of the output results of the early warning model when training the early warning model based on the complete accident data.

[0023] It should be noted that the above computer instructions may be stored in whole or in part on a computer-readable storage medium, wherein the computer-readable storage medium may be packaged together with the processor of the power grid data reconstruction device, or may be packaged separately from the processor of the power grid data reconstruction device, which is not limited in this application.

[0024] The description of the second, third, fourth and fifth aspects of the present application can refer to the detailed description of the first aspect; and the beneficial effects of the description of the second, third, fourth and fifth aspects can refer to the beneficial effect analysis of the first aspect, which will not be repeated here.

[0025] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description.

[0026] It is understandable that before using the technical solutions disclosed in the embodiments of this application, the type, scope of use, and usage scenarios of the personal information involved in this application should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0028] Figure 1 It is a flowchart of a method for reconstructing power grid data provided by an embodiment of the present application;

[0029] Figure 2 is a structural schematic diagram of a power grid data reconstruction device provided in an embodiment of the present application;

[0030] Figure 3 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the present application will be clearly and completely described below in conjunction with the drawings in the present embodiment. Obviously, the described embodiment is only a part of the embodiment of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present application.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only the parts related to the present application, rather than all structures, are shown in the accompanying drawings.

[0034] Figure 1 A flowchart of a power grid data reconstruction method provided in an embodiment of the present application is provided. This embodiment is applicable to the case where actual accident data is reconstructed before training an early warning model related to a power grid accident. The method can be performed by a power grid data reconstruction device, which can be implemented in the form of hardware and / or software and integrated in an electronic device that executes the method. Preferably, the electronic device in the embodiment of the present application can be a server, or a computer device, etc.

[0035] refer to Figure 1 The power grid data reconstruction method of this embodiment includes but is not limited to the following steps:

[0036] S110. Construct a sample data set according to the collected power grid accident information, where the power grid accident information includes at least two accident dimensions, and each accident dimension includes a plurality of accident data.

[0037] The solution provided in this embodiment is suitable for analyzing historical accident data corresponding to historical power grid accidents, so as to train an early warning model through historical accident data, so as to achieve the purpose of safety early warning of power grid accidents. Among them, historical power grid accidents can be: accidents in which power grid line failures are caused by falling trees or electric poles in natural environments; accidents in which casualties are caused by line problems during maintenance by staff; or accidents in which users cannot live normally due to sudden power outages, or property losses are caused in public areas, etc. The historical power grid data may include environmental data corresponding to the occurrence of power grid accidents, equipment operation data, accident type data, accident cause data, accident scope data, and accident loss data, etc.

[0038] In order to collect complete accident data, this embodiment needs to construct a sample data set based on the collected power grid accident information in advance. Among them, the power grid accident information represents the information corresponding to the power grid accident when the loss is caused by the power grid failure. Generally, the current accident information can be text information published by power grid enterprises or news agencies on relevant websites, such as relevant reports / explanations on a certain power grid accident, etc.; it can also be equipment operation information obtained from the power grid system when the power grid accident occurs.

[0039] The power grid accident information obtained in this embodiment includes at least two accident dimensions. In this embodiment, the accident dimension can be a dimension obtained by dividing according to factors such as accident cause, accident impact range, accident occurrence time, accident level and accident type, and the specific way of dividing the accident dimension is not limited here.

[0040] Among them, taking the dimension of accident causes as an example, this dimension may include equipment failure (transformer failure, circuit breaker failure, line failure or other equipment failure), natural disasters (lightning, wind disasters, fire, flood or earthquake) and human factors (misoperation, external damage or theft); taking the dimension of accident impact range as an example, this dimension may include local accidents (accidents affecting a single device or a local area, such as a transformer failure, a line tripping), regional accidents (affecting the operation of the power grid in a certain area, such as a substation power outage, multiple line failures in a region, etc.) and large-scale power outage accidents (affecting a larger area of ​​the power grid, and even causing power outages in the entire region or city); taking the dimension of accident occurrence time as an example, this dimension may include peak period accidents (accidents occurring during peak power consumption periods, when the power grid load is large and the accident may have a greater impact on power supply reliability), valley period accidents (accidents occurring during valley power consumption periods, when the power grid load is small and the impact of the accident is relatively small) and special period accidents. Therefore, (such as accidents occurring during special periods such as holidays and major events, which have higher requirements for power supply reliability); taking the accident level dimension as an example, this dimension can include general accidents (accidents that have little impact on the safe and stable operation of the power grid, causing small economic losses and social impacts), major accidents (accidents that have a certain impact on the safe and stable operation of the power grid, causing certain economic losses and social impacts), major accidents (accidents that have a greater impact on the safe and stable operation of the power grid, causing greater economic losses and social impacts) and particularly major accidents (accidents that have a serious impact on the safe and stable operation of the power grid, causing major economic losses and social impacts); taking the accident type dimension as an example, this dimension can include short-circuit accidents (including three-phase short circuits, two-phase short circuits, single-phase grounding short circuits, etc.), grounding accidents (system grounding faults, which may cause equipment damage and power outages), overvoltage accidents (including operating overvoltage, lightning overvoltage, etc.) and low voltage accidents (system voltage is too low, which may affect the normal operation of equipment and user power consumption). By dividing power grid accidents into multiple accident dimensions, we can better understand the characteristics and laws of accidents and provide a basis for accident analysis, prevention and handling. The accident classification contained in each accident dimension is not limited here.

[0041] Each accident dimension includes a plurality of corresponding accident data, and the current accident data may be related data that can be divided into accident dimensions. For example, in the accident text, it is described that "the cause of this accident is a large-scale power outage caused by a transformer failure", so "transformer failure" and "large-scale power outage" can be understood as accident data, and then the accident dimension can be determined as the accident cause dimension according to "transformer failure", and the accident dimension can be determined as the accident impact range dimension according to "large-scale power outage", etc.

[0042] Optionally, when the cause of the accident is an equipment failure and the power grid accident information includes the accident time and accident location, the corresponding accident power grid system can be determined according to the accident location, so as to obtain the equipment's power monitoring data, accident report data and operation and maintenance logs from the power grid accident power grid system according to the accident time, thereby facilitating further analysis of the specific causes of the equipment failure, such as winding short circuit, core failure, insulation damage, etc. The content of the accident data contained in each specific accident dimension is not limited here.

[0043] In a preferred embodiment, in this embodiment, the method of constructing a sample data set based on the collected power grid accident information can be implemented by the following steps:

[0044] a) Obtaining accident text related to power grid accidents from at least one type of web page addresses according to a preset corpus.

[0045] The preset corpus indicates the corpus of keywords / key sentences related to power grid accidents, so as to screen out relevant accident texts that preliminarily meet the requirements of power grid accidents through the preset corpus. For example, the preset corpus can be "power grid", "power grid accident", "power grid equipment", "power outage accident" or "power grid equipment failure", etc. Optionally, the preset corpus can include multiple corpora at the same time, so as to obtain corresponding accident texts from different angles based on multiple preset corpora, etc., and the determination of specific preset corpora is not limited here.

[0046] At least one type of web page address includes but is not limited to news media websites, power industry forums, government department announcement websites, search engine websites, power grid enterprise websites, and energy enterprise websites. Different types of web page addresses may provide power grid accident texts from different angles and with different levels of detail.

[0047] In this embodiment, web crawler technology or other data collection tools can be used to search and extract accident texts related to power grid accidents from various web page addresses according to preset corpus. The accident texts in this embodiment may include news reports, forum posts, announcements, etc. In the acquisition process, it is necessary to remove irrelevant information, advertisements and other interfering content to ensure that the extracted accident texts are directly related to power grid accidents.

[0048] b) Classify the text content in the accident text based on the accident dimension to obtain the accident data corresponding to each accident dimension.

[0049] In this embodiment, the accident dimensions may be preset, such as the accident cause dimension, accident impact scope dimension, accident occurrence time dimension, accident level dimension, and accident type dimension, etc. In addition, corresponding preset accident data is determined for each accident dimension, so that it is convenient to classify according to the text content in the obtained accident text, so that the accident data corresponding to each accident dimension can be obtained from the currently obtained text content.

[0050] Optionally, in this embodiment, the text content in the accident text is classified based on the accident dimension, and the accident data corresponding to each accident dimension can be obtained by using natural language processing technology, such as text classification algorithm, keyword extraction, etc. to determine the accident dimension to which the text belongs. For example, by analyzing the keywords and description content in the text, a news report about a power outage caused by a transformer failure in a certain area can be classified into accident dimensions such as "equipment failure" and "accident location".

[0051] After classification, each accident dimension corresponds to a set of accident data. These data may include text descriptions, relevant time, location, impact range, etc. The accident data of each accident dimension is sorted and stored to facilitate the subsequent construction of a sample data set.

[0052] c) Construct a sample data set based on the accident data corresponding to each accident dimension.

[0053] Further organize and clean the accident data corresponding to each accident dimension. Remove duplicate data, correct erroneous information, unify data formats, etc. to ensure data quality and consistency. Construct a sample data set based on the organized accident data. The sample data set can be stored in the form of tables, databases, etc., so that it can be called in subsequent model training.

[0054] The method of constructing a sample data set provided in this embodiment can obtain accident text related to power grid accidents from web page addresses, and classify them based on accident dimensions to construct a sample data set for analyzing and studying power grid accidents. In practical applications, the preset corpus, accident dimensions, and data processing methods can be continuously optimized according to specific needs to improve the quality and practicality of the sample data set.

[0055] In another preferred embodiment, in the above embodiment, when constructing a sample data set according to the accident data corresponding to each accident dimension, there may be a small amount of data samples. Specifically, the solution provided in this embodiment can be implemented in the following way:

[0056] Determine the number of fields in the accident data corresponding to each accident dimension; enhance the corresponding sample data according to the number of fields corresponding to each accident dimension to obtain enhanced sample data corresponding to each accident dimension; construct a sample data set according to the accident text of each accident dimension and the corresponding enhanced sample data. The purpose of this is that when the sample data set constructed by the method provided in this embodiment is used for subsequent power grid accident analysis, model training and decision support, it can improve the understanding and response capabilities of power grid accidents.

[0057] The accident field is used to represent the fields related to the current accident dimension. The accident fields corresponding to each accident dimension are preset. For example, for the accident type dimension, possible accident fields include "short circuit fault", "ground fault" or "equipment fault type" and other related descriptions; for the accident location dimension, possible accident fields include "province", "city" or "specific address" and other related descriptions.

[0058] For each accident dimension, all accident data are traversed, and the number of occurrences of each accident field corresponding to the dimension in the accident data is counted. By counting, the number of fields in the accident data corresponding to each accident dimension can be determined. For example, if the "specific equipment model" field under a certain accident dimension appears 20 times in 100 accident data, then the number of this field is 20.

[0059] In this embodiment, the accident fields whose number of fields is less than the preset number can be enhanced, so as to effectively increase the number of samples of the minority categories in the data set by sampling the minority category samples multiple times, thereby alleviating the category imbalance problem and providing a more balanced data basis for subsequent model training. The current preset number can be 20 or 50, etc., and the selection of the specific preset number is not limited here, and is subject to the actual sample requirements during model training.

[0060] Specifically, the corresponding sample data is enhanced according to the number of fields corresponding to each accident dimension. The enhanced sample data corresponding to each accident dimension can be obtained by: enhancing the data by random sampling and expansion. For example, for the dimension of the accident location, some locations can be randomly selected and new sample data can be generated in combination with other accident features; or for samples with a small number, data can be copied and the copied data can be changed to synonyms, similar words, etc. to increase the diversity of data samples. The specific method of enhancing the sample data is not limited here.

[0061] S120. Perform feature extraction on the accident data corresponding to at least two accident dimensions to obtain accident features corresponding to each accident dimension.

[0062] In this embodiment, feature extraction can be performed on the accident data corresponding to each accident dimension based on a character-level convolutional neural network (Char-CNN) to convert the text data corresponding to the accident data into a character-level or word-level vector for representation, wherein the character-level representation can capture the fine-grained information features of the text, while the word-level representation can capture the semantic information features of the text, which helps to effectively capture the semantic relationship in the accident data.

[0063] Optionally, different feature extraction methods can be selected for different accident dimensions. For example, for the accident type dimension, keywords, themes and other features can be extracted from the accident description text. Use text processing techniques, such as word frequency statistics or keyword extraction algorithms, to determine the key feature words related to different accident types; for the accident location dimension, the location information can be encoded or geographic features can be extracted. For example, geographic coordinates can be used to represent the location, or climate, topography and other features related to the location can be extracted; for the accident impact range dimension, statistical features can be extracted based on numerical data such as the number of affected users and the area of ​​the power outage area, etc. The specific method of extracting features for different accident dimensions is not limited here.

[0064] The purpose of obtaining the accident features corresponding to each accident dimension in the current step is to use the accident features obtained by analysis as input for the training model in the subsequent step S130, so that the model can better understand and predict the relationship between the accident data.

[0065] In a preferred implementation, in this embodiment, the above step S120 can be implemented by the following steps:

[0066] d) Perform sequence processing on each accident data to obtain the character sequence corresponding to each accident data.

[0067] Before performing sequence processing on each accident data, the accident data is first preprocessed. The current preprocessing operation can include removing special characters, punctuation marks, and stop words to reduce noise and irrelevant information; the text can also be segmented to divide the sentence into words for better subsequent sequence processing.

[0068] When obtaining the character sequence corresponding to each accident data, each character is converted into a corresponding digital representation according to the encoding method. Exemplarily, the current character encoding method can be implemented by ASCII code. In this embodiment, each accident data corresponds to a character sequence, which contains relevant information of the accident data. For example, when the accident data is a description text, each character in the text can be connected in sequence to form a character sequence.

[0069] e) Based on the convolutional neural network model, a convolution operation is performed on the character sequence corresponding to each accident data to obtain the local features corresponding to each accident data.

[0070] In the convolutional neural network model, a convolution kernel is used to perform a convolution operation on the input character sequence. The convolution kernel slides over the character sequence and performs a convolution operation with a local character window to extract local features. The size and number of convolution kernels can be adjusted according to task requirements. Smaller convolution kernels can extract local character patterns, while larger convolution kernels can capture more extensive contextual information.

[0071] f) Determine the accident characteristics of the corresponding accident dimension according to the local characteristics corresponding to each accident data.

[0072] When determining the accident features corresponding to the accident dimension, the local features corresponding to each accident data contained in the current accident dimension can be processed by feature fusion, thereby integrating the features extracted by different convolution kernels to improve the expressiveness of the features.

[0073] Optionally, the local features corresponding to each accident data may be fused by concatenation, summation or weighted summation to obtain accident features corresponding to the accident dimension.

[0074] S130. Training a preset network model based on the accident features corresponding to each accident dimension to obtain a data reconstruction model.

[0075] In this embodiment, the preset network model can be a long short-term memory network (Long Short-Term Memory, LSTM for short). The LSTM network can capture the long-term dependencies and context information in the accident characteristics corresponding to each accident dimension, such as being able to analyze the association between the accident cause dimension and the accident type dimension, and perform multi-dimensional analysis of the power grid accident data based on the LSTM network, such as time series analysis, spatial distribution analysis, etc., to analyze the patterns between the accident data of the power grid accidents that occur.

[0076] In the process of training the preset network model based on the accident features corresponding to each accident dimension to obtain the data reconstruction model, the data samples in the sample data set can be divided into a training set, a validation set and a test set in advance according to a certain ratio, and the accident features corresponding to each accident dimension in the training set are input into the preset network model. The preset network model learns the dependency relationship between the accident features and continuously adjusts the model parameters in the preset network model to improve the prediction accuracy of the model until the model converges or reaches a predetermined number of training times to obtain the data reconstruction model.

[0077] Another preferred implementation method, in this embodiment, the above step S130 can be specifically implemented in the following way: determine the dependency relationship between every two accident dimensions based on the accident characteristics corresponding to each accident dimension; train the preset network model according to the dependency relationship between every two accident dimensions, and obtain the data reconstruction model when the training result reaches convergence.

[0078] The significance of determining the dependency between every two accident dimensions based on the accident characteristics corresponding to each accident dimension is that it is found that there is a certain correlation between the time of accident and the type of accident, which may mean that certain types of accidents are more likely to occur in a specific time period; it can also be found that specific weather conditions in a certain area always appear before certain types of accidents, so it can be inferred that weather conditions may be a cause of the accident, etc.

[0079] Specifically, the method for determining the dependency relationship between every two accident dimensions can be to calculate the linear or nonlinear relationship between the two accident dimensions through the Wilson correlation coefficient or the Spearman rank correlation coefficient; or the dependency relationship can be modeled, such as constructing a Bayesian network, in which each node represents an accident dimension, the edge between the nodes represents the dependency relationship, the weight of the edge represents the strength of the dependency, etc. The specific method for determining the dependency relationship between every two accident dimensions is not limited here.

[0080] When training the preset network model according to the dependency between every two accident dimensions, if there is a strong dependency between the two accident dimensions, corresponding connections can be added to the model or weights can be adjusted to better capture this relationship, and the model parameters are continuously adjusted and optimized through the determined dependency until the loss function reaches a minimum or the number of training times reaches an upper limit, and a data reconstruction model is obtained. The data reconstruction model in this embodiment utilizes the dependency between accident dimensions, can better capture the complexity and correlation of data, and improve the performance and generalization ability of the model.

[0081] S140: Input the accident text corresponding to the target power grid accident into the data reconstruction model to obtain the missing data of the target power grid accident.

[0082] The target power grid accident represents a power grid accident that occurred in a historical period, but some data in this power grid accident is missing, such as the accident loss and accident level. Based on this, the accident text corresponding to the target power grid accident can be input into the data reconstruction model, and the known accident data and missing data can be extracted from the accident text corresponding to the target power grid accident through the data reconstruction model, and then the confirmed data can be predicted based on the data reconstruction model and the known accident data, so as to achieve complete reconstruction of the power grid accident data.

[0083] In another preferred embodiment, in this embodiment, the above step S140 can be implemented as follows:

[0084] Acquire known accident data corresponding to known accident dimensions from the accident text; determine the accident dimensions to be reconstructed based on the known accident dimensions; input the known accident data corresponding to the known accident dimensions and the accident dimensions to be reconstructed into the data reconstruction model to obtain the missing data corresponding to the accident dimensions to be reconstructed.

[0085] Specifically, the method of reconstructing the missing data of the target power grid accident according to the trained data reconstruction model is implemented in the following way:

[0086] First, the accident text is processed using natural language processing technology, such as text segmentation, part-of-speech tagging, named entity recognition, etc., to extract the known accident data in the text and determine the known accident dimensions corresponding to the known accident data. For example, the current known accident data can be the relevant accident data of the accident dimensions such as the time, location, and type of accident that are identified from the accident text; then, according to the needs of the actual application, the accident dimension that needs to be reconstructed is determined. This may be because the data of this dimension is missing, inaccurate, or needs to be further refined. For example, if the impact range of the accident needs to be predicted, but the data of this dimension is not directly provided in the accident text, then the impact range can be determined as the accident dimension to be reconstructed; then the known accident data corresponding to the known accident dimension is sorted and formatted to meet the input requirements of the data reconstruction model, and the accident dimension to be reconstructed is input as the output target of the model. Specific identifiers or placeholders can be used to represent the dimension to be reconstructed; finally, the data reconstruction model processes the input data and outputs the predicted value of the accident dimension to be reconstructed. This predicted value is the missing data corresponding to the accident dimension to be reconstructed. Through the solution provided in this embodiment, existing data and models can be effectively utilized to improve the integrity and accuracy of power grid accident data, provide better support for accident analysis and management, and lay the foundation for subsequent training of early warning models related to power grid accidents.

[0087] The power grid data reconstruction method provided in this embodiment first constructs a sample data set based on the collected power grid accident information, where the power grid accident information includes at least two accident dimensions, and each accident dimension includes multiple accident data; then, feature extraction is performed on the accident data corresponding to at least two accident dimensions to obtain the accident features corresponding to each accident dimension; then, the preset network model is trained based on the accident features corresponding to each accident dimension to obtain a data reconstruction model; finally, the accident text corresponding to the target power grid accident is input into the data reconstruction model to obtain the missing data of the target power grid accident. The data reconstruction model constructed by this scheme can learn the accident data corresponding to multiple accident dimensions, so that the data reconstruction model completed through training can output the missing data corresponding to any power grid accident to ensure the integrity of the accident data, thereby helping to improve the accuracy and reliability of the output results of the early warning model when training the early warning model based on the complete accident data.

[0088] Figure 2 is a schematic diagram of a structure of a power grid data reconstruction device provided in an embodiment of the present application, and the device is suitable for executing the power grid data reconstruction method provided in an embodiment of the present application. Figure 2As shown, the device may specifically include: a data set construction module 210, a feature extraction module 220, a model training module 230 and a data reconstruction module 240, wherein:

[0089] A data set construction module 210 is used to construct a sample data set according to the collected power grid accident information, wherein the power grid accident information includes at least two accident dimensions, and each of the accident dimensions includes a plurality of accident data;

[0090] A feature extraction module 220 is used to extract features from the accident data corresponding to at least two of the accident dimensions to obtain an accident feature corresponding to each of the accident dimensions;

[0091] A model training model 230, used to train a preset network model based on the accident features corresponding to each of the accident dimensions to obtain a data reconstruction model;

[0092] The data reconstruction module 240 is used to input the accident text corresponding to the target power grid accident into the data reconstruction model to obtain the missing data of the target power grid accident.

[0093] The power grid data reconstruction device provided in this embodiment first constructs a sample data set based on the collected power grid accident information, where the power grid accident information includes at least two accident dimensions, and each accident dimension includes multiple accident data; then, feature extraction is performed on the accident data corresponding to at least two accident dimensions to obtain the accident features corresponding to each accident dimension; then, the preset network model is trained based on the accident features corresponding to each accident dimension to obtain a data reconstruction model; finally, the accident text corresponding to the target power grid accident is input into the data reconstruction model to obtain the missing data of the target power grid accident. The data reconstruction model constructed by this scheme can learn the accident data corresponding to multiple accident dimensions, so that the data reconstruction model completed through training can output the missing data corresponding to any power grid accident to ensure the integrity of the accident data, thereby helping to improve the accuracy and reliability of the output results of the early warning model when training the early warning model based on the complete accident data.

[0094] In one embodiment, the data set construction module 210 includes a text acquisition unit, a text classification unit and a data set construction unit, wherein:

[0095] A text acquisition unit, used for acquiring accident text related to power grid accidents from at least one type of web page address according to a preset corpus;

[0096] A text classification unit, used to classify the text content in the accident text based on the accident dimension, and obtain accident data corresponding to each accident dimension;

[0097] The data set construction unit is used to construct the sample data set according to the accident data corresponding to each accident dimension.

[0098] In one embodiment, the data set construction unit is specifically used to determine the number of fields in the accident data corresponding to each accident dimension of the accident field; enhance the corresponding sample data according to the number of fields corresponding to each accident dimension to obtain enhanced sample data corresponding to each accident dimension; and construct the sample data set according to the accident text of each accident dimension and the corresponding enhanced sample data.

[0099] In one embodiment, the feature extraction module 220 includes a sequence processing unit, a convolution operation unit and a feature determination unit, wherein:

[0100] A sequence processing sheet, used to perform sequence processing on each of the accident data to obtain a character sequence corresponding to each of the accident data;

[0101] A convolution operation unit, used for performing a convolution operation on a character sequence corresponding to each of the accident data based on a convolutional neural network model to obtain a local feature corresponding to each of the accident data;

[0102] The feature determination unit is used to determine the accident feature corresponding to the accident dimension according to the local feature corresponding to each of the accident data.

[0103] In one embodiment, the model training model 230 is specifically used to determine the dependency relationship between every two accident dimensions based on the accident characteristics corresponding to each of the accident dimensions; the preset network model is trained according to the dependency relationship between every two accident dimensions, and when the training results reach convergence, the data reconstruction model is obtained.

[0104] In one embodiment, the data reconstruction module 240 includes a known data acquisition unit, a data dimension determination unit, and a missing data acquisition unit, wherein:

[0105] A known data acquisition unit, used to acquire known accident data corresponding to a known accident dimension from the accident text;

[0106] A data dimension determination unit, used to determine the accident dimension to be reconstructed according to the known accident dimension;

[0107] The missing data obtaining unit is used to input the known accident data corresponding to the known accident dimension and the accident dimension to be reconstructed into the data reconstruction model to obtain the missing data corresponding to the accident dimension to be reconstructed.

[0108] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the functional modules described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0109] An embodiment of the present application also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the power grid data reconstruction method described in any embodiment of the present application.

[0110] An embodiment of the present application further provides a computer-readable medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the power grid data reconstruction method described in any embodiment of the present application when executed.

[0111] Reference below Figure 3 , Figure 3 1 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application, and shows a schematic diagram of the structure of a computer system 500 suitable for implementing the electronic device in an embodiment of the present application. Figure 3 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0112] like Figure 3 As shown, the computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage part 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the system 500 are also stored. The CPU 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0113] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed, so that a computer program read therefrom is installed into the storage section 508 as needed.

[0114] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 509, and / or installed from the removable medium 511. When the computer program is executed by the central processing unit (CPU) 501, the above-mentioned functions defined in the system of the present application are executed.

[0115] It should be noted that the computer-readable medium shown in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, etc., or any suitable combination of the above.

[0116] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the above-mentioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0117] The modules and / or units involved in the embodiments described in the present application may be implemented by software or by hardware. The modules and / or units described may also be provided in a processor, for example, may be described as: a processor including a data set construction module, a feature extraction module, and a model training model. The names of these modules do not, in some cases, constitute limitations on the modules themselves.

[0118] As another aspect, the present application also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently and not be assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device includes: constructing a sample data set based on the collected power grid accident information, the power grid accident information includes at least two accident dimensions, and each of the accident dimensions includes multiple accident data; performing feature extraction on the accident data corresponding to at least two of the accident dimensions to obtain the accident features corresponding to each of the accident dimensions; training a preset network model based on the accident features corresponding to each of the accident dimensions to obtain a data reconstruction model; inputting the accident text corresponding to the target power grid accident into the data reconstruction model to obtain the missing data of the target power grid accident.

[0119] According to the technical solution of this embodiment, the data reconstruction model can learn the accident data corresponding to multiple accident dimensions, so that the data reconstruction model completed through training can output the missing data corresponding to any power grid accident to ensure the integrity of the accident data, thereby helping to improve the accuracy and reliability of the output results of the early warning model when training the early warning model based on the complete accident data.

[0120] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for reconstructing power grid data, characterized in that: include: Constructing a sample data set according to the collected power grid accident information, wherein the power grid accident information includes at least two accident dimensions, and each of the accident dimensions includes a plurality of accident data; Performing feature extraction on the accident data corresponding to at least two of the accident dimensions to obtain an accident feature corresponding to each of the accident dimensions; Training a preset network model based on the accident features corresponding to each of the accident dimensions to obtain a data reconstruction model; The accident text corresponding to the target power grid accident is input into the data reconstruction model to obtain the missing data of the target power grid accident.

2. The power grid data reconstruction method according to claim 1, characterized in that: The sample data set is constructed based on the collected power grid accident information, including: Acquire accident text related to power grid accidents from at least one type of web page addresses according to a preset corpus; Classifying the text content in the accident text based on the accident dimension to obtain accident data corresponding to each accident dimension; The sample data set is constructed according to the accident data corresponding to each of the accident dimensions.

3. The power grid data reconstruction method according to claim 2, characterized in that: The constructing the sample data set according to the accident data corresponding to each of the accident dimensions includes: Determine the number of fields of the accident field corresponding to each of the accident dimensions in the accident data; Performing enhancement processing on the corresponding sample data according to the number of fields corresponding to each of the accident dimensions, to obtain enhanced sample data corresponding to each of the accident dimensions; The sample data set is constructed according to the accident text of each accident dimension and the corresponding enhanced sample data.

4. The power grid data reconstruction method according to claim 1, characterized in that: The extracting features of the accident data corresponding to at least two of the accident dimensions to obtain the accident features corresponding to each of the accident dimensions includes: Performing sequence processing on each of the accident data to obtain a character sequence corresponding to each of the accident data; Performing a convolution operation on the character sequence corresponding to each of the accident data based on a convolutional neural network model to obtain a local feature corresponding to each of the accident data; The accident feature corresponding to the accident dimension is determined according to the local feature corresponding to each of the accident data.

5. The method for reconstructing power grid data according to claim 1, characterized in that: The step of training a preset network model based on the accident features corresponding to each of the accident dimensions to obtain a data reconstruction model includes: Determine the dependency relationship between every two accident dimensions based on the accident features corresponding to each of the accident dimensions; The preset network model is trained according to the dependency relationship between every two accident dimensions, and when the training result reaches convergence, the data reconstruction model is obtained.

6. The method for reconstructing power grid data according to claim 1, characterized in that: The step of inputting the accident text corresponding to the target power grid accident into the data reconstruction model to obtain the missing data of the target power grid accident includes: Acquire known accident data corresponding to known accident dimensions from the accident text; Determining the accident dimension to be reconstructed according to the known accident dimension; The known accident data corresponding to the known accident dimension and the accident dimension to be reconstructed are input into the data reconstruction model to obtain the missing data corresponding to the accident dimension to be reconstructed.

7. A power grid data reconstruction device, characterized in that: include: A data set construction module, used to construct a sample data set according to the collected power grid accident information, wherein the power grid accident information includes at least two accident dimensions, and each of the accident dimensions includes a plurality of accident data; A feature extraction module, used to extract features from the accident data corresponding to at least two of the accident dimensions, to obtain accident features corresponding to each of the accident dimensions; A model training model, used to train a preset network model based on the accident features corresponding to each of the accident dimensions to obtain a data reconstruction model; The data reconstruction module is used to input the accident text corresponding to the target power grid accident into the data reconstruction model to obtain the missing data of the target power grid accident.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the power grid data reconstruction method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the power grid data reconstruction method as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the power grid data reconstruction method according to any one of claims 1 to 6.

Images