An abnormality reporting information generation method and device of a power distribution network, a terminal device, and a storage medium

Abstract

The application discloses a power distribution network abnormality report information generation method and device, a terminal device and a storage medium. The application can convert power grid abnormality information into corresponding semantic vectors, thereby obtaining a corresponding semantic vector list. The similarity of each semantic vector in the semantic vector list and an index vector of each regulation document is calculated. The regulation document with the highest similarity is taken as a target regulation document that is most matched and most similar to the power grid abnormality information, without manual checking and comparison. The reporting content in the target regulation document can also be automatically extracted. Finally, based on the reporting content and each abnormality text of the power grid abnormality information, abnormality report information corresponding to the power grid abnormality information is automatically generated. The automatic matching of the regulation document is realized, and the reporting content in the target regulation document can be automatically extracted, avoiding errors caused by human factors. The reporting efficiency is improved, and the accuracy of the abnormality report information is improved.

Description

Technical Field

[0001] This invention relates to the field of computer processing technology, and in particular to a method, apparatus, terminal device, and storage medium for generating abnormal reporting information in a power distribution network. Background Technology

[0002] In modern power grid dispatching, the rapid and accurate reporting of grid anomalies has become crucial for ensuring grid event synchronization and maintaining stable grid operation. Anomalies may involve equipment failures, safety hazards, operational errors, and other aspects. Timely and effective reporting helps organizations mitigate potential risks and respond swiftly. Procedure documents typically detail the handling processes and reporting requirements for various anomalies. Consulting these documents helps dispatchers accurately determine the anomaly level, the reporting unit, and specific reporting requirements. This allows them to determine the corresponding anomaly reporting information, ensuring that subsequent anomaly reports are accurately delivered to the appropriate unit.

[0003] However, in the traditional abnormal information reporting process, it usually relies on manual review of a large number of procedural documents to match the corresponding procedural documents, and on the personnel's experience to extract the necessary abnormal situation handling procedures and reporting requirements, such as the reporting unit, the reporting time, or the reporting method. The traditional abnormal information reporting method relies on manual matching of procedural documents, which consumes a lot of time and energy, resulting in a low generation rate of abnormal reporting information. It is also prone to errors due to personal negligence or misunderstanding, affecting the accuracy of the abnormal reporting information. Summary of the Invention

[0004] This invention provides a method, apparatus, terminal device, and storage medium for generating abnormal reporting information in a power distribution network. It can convert each abnormal text in the abnormal information of the power grid into a semantic vector and perform similarity calculation with the index vector of the procedure document, thereby realizing the automated matching of the procedure document. It can effectively solve the problem that the generation rate and accuracy of abnormal reporting information are low due to the reliance on manual review of a large number of procedure documents to match the corresponding procedure documents in the prior art.

[0005] An embodiment of the present invention provides a method for generating anomaly reporting information in a distribution network, comprising:

[0006] Obtain power grid anomaly information input by the user; wherein, the power grid anomaly information includes: several anomaly texts used to describe the anomaly phenomenon;

[0007] The power grid anomaly information is converted into a semantic vector list. Based on the semantic vectors in the semantic vector list and the index vector of each procedure document, the similarity between each procedure document and the semantic vector list is calculated. Each semantic vector in the semantic vector list corresponds to an anomaly text. The procedure document is used to represent the reporting rules corresponding to the handling of power grid anomalies.

[0008] The procedure document with the highest similarity is selected as the target procedure document, and then the corresponding reporting content in the target procedure document is extracted; wherein, the reporting content includes: the reporting unit, the reporting time, and the reporting method;

[0009] Based on each abnormal text and the reported content, generate abnormal reporting information corresponding to the power grid abnormal information.

[0010] Preferably, the step of calculating the similarity between each procedure document and the semantic vector list based on each semantic vector in the semantic vector list and the index vector of each procedure document includes:

[0011] For each procedure document, the similarity between each semantic vector in the semantic vector list and the index vector of the procedure document is calculated to obtain the similarity between each semantic vector and the procedure document.

[0012] For each procedure document, the average similarity between the procedure document and the semantic vector list is obtained based on the similarity between each semantic vector in the semantic vector list and the procedure document, as well as the total number of semantic vectors in the semantic vector list; the average similarity is used as the similarity between the procedure document and the semantic vector list.

[0013] Preferably, the reported content further includes a reporting template;

[0014] The extraction of the corresponding reporting content from the target procedure document includes:

[0015] Obtain a preset prompt template for guiding the model to output preset content; wherein, the preset prompt template includes: reporting unit, reporting time, reporting method, and reporting template;

[0016] Each abnormal text in the power grid abnormal information is segmented into words, and a word segmentation list corresponding to the power grid abnormal information is generated based on each segmented word.

[0017] The preset prompt template, target procedure document, and word segmentation list are input into a preset document processing model. The document processing model extracts word features of each word in the word segmentation list and captures contextual information of each text paragraph in the target procedure document. Based on the contextual information of each text paragraph, corresponding text features are generated. Based on each word feature and each text feature, the text paragraph with the highest relevance to the word segmentation list is extracted from the target procedure document. Based on the preset prompt template and the text paragraph, the reporting content corresponding to the preset prompt template is generated.

[0018] The generation of the preset document processing model includes:

[0019] Training samples are generated based on the word segmentation list samples corresponding to the procedure document samples, the power grid anomaly information samples, and the preset prompt template samples; wherein, the preset prompt template samples include: reporting unit, reporting time, reporting method, and reporting template;

[0020] The document processing model to be trained is iteratively trained using each training sample and the actual reported content corresponding to each training sample as input, and the predicted reported content of each training sample as output, until the model converges, thus generating the preset document processing model.

[0021] Preferably, during each iteration of training, a training sample is input into the document processing model so that the document processing model can extract the text paragraph with the highest relevance to the word segmentation list sample from the procedure document sample based on the word features and text features in the training sample, and generate the reporting content prediction result corresponding to the preset prompt template sample based on the preset prompt template and the text paragraph.

[0022] The predicted results of the reported content are compared with the actual reported content, and the network parameters of the document processing model are adjusted based on the comparison results.

[0023] Preferably, the document processing model includes: an attention mechanism layer;

[0024] The step of extracting the text paragraphs with the highest relevance to the word segmentation list from the target procedure document based on each word feature and each text feature includes:

[0025] The attention mechanism layer is used to calculate the cosine similarity between each word feature and each text feature.

[0026] For each text feature, based on the activation function, the cosine similarity between the text feature and each word feature is converted into a probability distribution, and attention weights are assigned to the text features according to each probability distribution; wherein, the attention weights are used to characterize the degree of correlation between the text features and the word segmentation list;

[0027] Based on each attention weight, the text paragraph corresponding to the text feature with the highest attention weight is determined.

[0028] Preferably, the document processing model includes: a fusion layer;

[0029] The step of generating reporting content corresponding to the preset prompt template based on the preset prompt template and the text paragraph includes:

[0030] The fusion layer is used to concatenate the first text vectors corresponding to the preset prompt template and the second text vectors corresponding to the text paragraphs in the vector space to generate several concatenated third text vectors; wherein, the first text vectors corresponding to the preset prompt template include: unit text vector, time text vector, method text vector, and template text vector; the second text vectors corresponding to the text paragraphs include: reporting text vector, reporting time text vector, reporting method text vector, and reporting template text vector;

[0031] Based on each concatenated third text vector, generate the reporting content corresponding to the preset prompt template.

[0032] Preferably, the abnormal text includes: the location of the abnormality, the time of the abnormality, the type of the abnormal device, and the abnormal status; the abnormal status is used to characterize the device status when the abnormality occurs; the reporting template includes: a time filling part, a unit filling part, a device type filling part, and an abnormal status filling part;

[0033] The step of generating abnormal reporting information corresponding to the power grid abnormal information based on each abnormal text and the reported content includes:

[0034] Fill the time of the anomaly, the reporting unit, the type of the abnormal device, and the state of the anomaly into the time, unit, device type, and anomaly situation fields of the reporting template respectively to generate a filled reporting template.

[0035] Based on the filled-in reporting template, the abnormal text of the abnormal location, the reporting time in the reporting content, and the reporting method in the reporting content, the abnormal reporting information corresponding to the power grid abnormal information is generated.

[0036] Based on the above method embodiments, the present invention provides corresponding apparatus embodiments.

[0037] An embodiment of the present invention provides an anomaly reporting information generation device for a power distribution network, comprising: an anomaly information acquisition module, a similarity calculation module, a reporting content extraction module, and an anomaly reporting information generation module;

[0038] The abnormal information acquisition module is used to acquire power grid abnormal information input by the user; wherein, the power grid abnormal information includes: a number of abnormal texts used to describe abnormal phenomena;

[0039] The similarity calculation module is used to convert the power grid anomaly information into a semantic vector list, and calculate the similarity between each procedure document and the semantic vector list based on each semantic vector in the semantic vector list and the index vector of each procedure document; wherein, each semantic vector in the semantic vector list corresponds to an anomaly text; the procedure document is used to represent the reporting rules corresponding to the handling of power grid anomalies.

[0040] The reporting content extraction module is used to select the procedure document with the highest similarity as the target procedure document, and then extract the corresponding reporting content in the target procedure document; wherein, the reporting content includes: the reporting unit, the reporting time, and the reporting method;

[0041] The abnormal reporting information generation module is used to generate abnormal reporting information corresponding to the power grid abnormal information based on each abnormal text and the reporting content.

[0042] Based on the above method embodiments, the present invention provides corresponding terminal device embodiments.

[0043] Another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the method for generating abnormal reporting information of a power distribution network as described in the above-described embodiment of the invention.

[0044] Based on the above method embodiments, the present invention provides corresponding storage medium embodiments.

[0045] Another embodiment of the present invention provides a storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the method for generating abnormal reporting information of a power distribution network as described in the above-described embodiment of the invention.

[0046] The following benefits can be obtained by implementing the present invention:

[0047] This invention provides a method, apparatus, terminal device, and storage medium for generating abnormal reporting information in a power distribution network. Based on user-inputted power grid abnormal information, this invention can convert each abnormal text in the abnormal information into a corresponding semantic vector, thereby obtaining a semantic vector list corresponding to the abnormal information. Based on each semantic vector in this list and the index vector of each procedure document, the similarity between the semantic vector list and each procedure document can be calculated. Based on the procedure document with the highest similarity, the target procedure document most similar to the text of the abnormal information can be obtained without manual review and comparison. Furthermore, the reporting content in the target procedure document can be automatically extracted. Finally, based on the reporting unit, reporting time, and reporting method in the reporting content, and each abnormal text, the abnormal reporting information corresponding to the abnormal information in the power grid is automatically generated. Compared with existing technologies, this invention can convert each abnormal text in the power grid anomaly information into a semantic vector and perform similarity calculation with the index vector of the procedure document, thereby realizing automated matching of the procedure document. Furthermore, this invention improves reporting efficiency by automatically extracting the reporting content from the target procedure document. By avoiding errors caused by human factors, it also further improves the quality and accuracy of anomaly reporting information. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating a method for generating abnormal reporting information in a power distribution network according to an embodiment of the present invention.

[0049] Figure 2 This is a schematic diagram of the structure of a power distribution network anomaly reporting information generation device provided in an embodiment of the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] like Figure 1 The diagram shown is a flowchart illustrating a method for generating abnormal reporting information in a power distribution network according to an embodiment of the present invention. The method includes:

[0052] Step S1: Obtain power grid anomaly information input by the user; wherein, the power grid anomaly information includes: several anomaly texts used to describe the anomaly phenomenon;

[0053] Step S2: Convert the power grid anomaly information into a semantic vector list. Calculate the similarity between each procedure document and the semantic vector list based on the index vector of each semantic vector in the semantic vector list and the index vector of each procedure document. Each semantic vector in the semantic vector list corresponds to an anomaly text. The procedure document is used to characterize the reporting rules corresponding to handling power grid anomalies.

[0054] Step S3: Select the procedure document with the highest similarity as the target procedure document, and then extract the corresponding reporting content from the target procedure document; wherein, the reporting content includes: reporting unit, reporting time and reporting method;

[0055] Step S4: Generate abnormal reporting information corresponding to the power grid abnormal information based on each abnormal text and the reported content.

[0056] For step S1, in a preferred embodiment, the user can input abnormal text to describe in detail the abnormal situation that occurred in the power grid, such as the location of the abnormality, the time of the abnormality, the type of abnormal equipment, and the abnormal status. The abnormal status is used to characterize the equipment status when the abnormal situation occurs, such as whether the equipment is running, whether the operating parameters are abnormal (e.g., whether the current or voltage exceeds the normal range), whether the equipment has physical damage or fault indication, etc. The abnormal status describes the specific performance or condition of the equipment under abnormal conditions. Based on the abnormal status, the nature and severity of the abnormal event can be understood more comprehensively, thereby matching the corresponding procedure document.

[0057] In a preferred embodiment, for step S2, the present invention may convert the power grid anomaly information into a list of semantic vectors containing several semantic vectors before matching the power grid anomaly information with the procedure document, and the semantic vectors in the list correspond to the anomaly text.

[0058] Therefore, by converting power grid anomaly information into semantic vectors, the similarity of the vectors can be used to accurately match the procedure documents, thus capturing the semantic similarity between texts, rather than just the literal similarity, and improving the accuracy of document matching.

[0059] By calculating the similarity between each procedure document and the semantic vector list, the most relevant procedure document can be quickly located, reducing the time spent manually reviewing and filtering procedure documents and improving the efficiency of generating anomaly reporting information. Specifically, the calculation of the similarity between each procedure document and the semantic vector list includes:

[0060] For each procedure document, the similarity between each semantic vector in the semantic vector list and the index vector of the procedure document is calculated to obtain the similarity between each semantic vector and the procedure document.

[0061] For each procedure document, the average similarity between the procedure document and the semantic vector list is obtained based on the similarity between each semantic vector in the semantic vector list and the procedure document, as well as the total number of semantic vectors in the semantic vector list. This average similarity is then used as the corresponding similarity between the procedure document and the semantic vector list. It is understood that the average similarity can more comprehensively reflect the overall correlation between the procedure document and power grid anomaly information, and is an important basis for subsequent procedure document localization.

[0062] As an illustration, in order to match the semantic vector of power grid anomaly information, this invention also needs to construct an index vector for each procedure document. The constructed index vector can represent the main content and features of the procedure document, which is the basis for subsequent similarity calculation.

[0063] Procedure documents generally refer to standard documents that specify, explain, or guide a process, operation, or behavior. In the field of power dispatching, the procedure documents matched by this invention typically contain specific reporting rules, procedures, and requirements for handling abnormal power grid situations. These procedure documents serve as important references for dispatchers when dealing with power grid anomalies. Illustratively, a procedure document may include the following:

[0064] Reporting rules: The regulations document will clearly specify which unit(s) are responsible for reporting different types of power grid anomalies, as well as the priority and urgency of the reporting.

[0065] Reporting process: This section details the complete steps from the discovery of an anomaly to the final reporting, including information collection, analysis and judgment, decision-making, and reporting execution.

[0066] Reporting requirements include reporting time limits, reporting methods (such as written reports, telephone notifications, etc.), and the format and content requirements of the reported information.

[0067] Anomaly Handling Guidelines: In addition to reporting rules, the procedure documents may also include preliminary handling guidelines for abnormal power grid situations to help dispatchers take necessary emergency measures before reporting.

[0068] Relevant procedures and standards: Reference or include other procedures, standards or guidance documents related to power grid anomaly handling to ensure consistency and compliance of the reporting process.

[0069] This invention can automatically match relevant procedural documents based on the abnormal situation input by the dispatcher, thereby deriving the requirements and rules for abnormal reporting, and automatically generating abnormal reporting information for the current abnormal situation based on the relevant procedural documents.

[0070] As an illustration, even if the reporting rules for the procedure documents are updated, this invention only needs to directly update the content of the procedure documents in the knowledge base when the procedure documents change. There is no need for dispatchers to manually repeat the learning. Instead, the similarity calculation of each semantic vector in the power grid anomaly information with the index vector of the procedure documents can realize the automatic matching of the procedure documents, reducing the learning cost and time of the new procedure documents.

[0071] For step S3, in a preferred embodiment, the present invention can extract the procedure document with the highest similarity to the semantic vector list based on the similarity between each procedure document and the list of semantic vectors, and mark it as the target procedure document to be processed. Based on the target procedure document, the reporting content that can be used to generate abnormal reporting information can be extracted, such as the reporting unit, reporting time, reporting method and reporting template.

[0072] In illustrative terms, embodiments of the present invention can process and extract content related to anomaly reporting from standard procedure documents using a preset document processing model, resulting in:

[0073] Obtain a preset prompt template for guiding the model to output preset content; wherein, the preset prompt template includes: reporting unit, reporting time, reporting method, and reporting template;

[0074] Each abnormal text in the power grid abnormal information is segmented into words, and a word segmentation list corresponding to the power grid abnormal information is generated based on each segmented word.

[0075] The preset prompt template, target procedure document, and word segmentation list are input into a preset document processing model. The document processing model extracts word features of each word in the word segmentation list and captures contextual information of each text paragraph in the target procedure document. Based on the contextual information of each text paragraph, corresponding text features are generated. Based on each word feature and each text feature, the text paragraph with the highest relevance to the word segmentation list is extracted from the target procedure document. Based on the preset prompt template and the text paragraph, the reporting content corresponding to the preset prompt template is generated.

[0076] The generation of the preset document processing model includes:

[0077] Training samples are generated based on the word segmentation list samples corresponding to the procedure document samples, the power grid anomaly information samples, and the preset prompt template samples; wherein, the preset prompt template samples include: reporting unit, reporting time, reporting method, and reporting template;

[0078] The document processing model to be trained is iteratively trained using each training sample and the actual reported content corresponding to each training sample as input, and the predicted reported content of each training sample as output, until the model converges, thus generating the preset document processing model.

[0079] In illustratively, during each training iteration, a training sample is input into the document processing model so that the document processing model can extract the text paragraph with the highest relevance to the word segmentation list sample from the procedure document sample based on the word features and text features in the training sample, and generate the reporting content prediction result corresponding to the preset prompt template sample based on the preset prompt template and the text paragraph.

[0080] The predicted results of the reported content are compared with the actual reported content, and the network parameters of the document processing model are adjusted based on the comparison results.

[0081] Specifically, when processing documents using a trained document processing model, it is first necessary to obtain a preset prompt template to guide the model to output preset content. The preset prompt template is then input into the model, allowing the model to summarize and output the reporting unit, reporting time, reporting method, and reporting template based on the template. By using the preset prompt template, the model can extract and generate information in a more targeted and accurate manner.

[0082] In illustrative terms, the document processing model of the present invention includes: an input layer, a feature extraction layer, an attention mechanism layer, a fusion layer, and an output layer;

[0083] The input layer receives user-input error messages, target procedure documents, and preset prompt templates, converting the text information into a format that the model can process, such as converting text into word vectors. This ensures that different forms of input data can be processed uniformly, providing a foundation for subsequent feature extraction and model computation.

[0084] The feature extraction layer is used to extract features from each word in the word segmentation list, such as word frequency, part-of-speech tagging, and semantic vectors. It can also utilize structures such as Recurrent Neural Networks (RNNs), Long Short-Term Memory Networks (LSTMs), or Transformers to capture contextual information from each text paragraph in the target procedure document, generating corresponding text features. Thus, the feature extraction layer can extract rich text features, providing important information input for the subsequent attention mechanism layer.

[0085] The attention mechanism layer is used to calculate the cosine similarity between each word feature and each text feature.

[0086] For each text feature, based on the activation function, the cosine similarity between the text feature and each word feature is converted into a probability distribution, and attention weights are assigned to the text features according to each probability distribution; wherein, the attention weights are used to characterize the degree of correlation between the text features and the word segmentation list;

[0087] Based on each attention weight, the text paragraph corresponding to the text feature with the highest attention weight is determined.

[0088] The attention mechanism layer of this invention can calculate the correlation or similarity between word features and text features, and then extract the text paragraphs with the highest relevance to the word segmentation list from the target procedure document. Thus, the attention mechanism layer can extract the text paragraphs with the highest relevance to the word segmentation list (i.e., the abnormal information input by the user), thereby ensuring that the information processed subsequently is accurate and targeted. In other words, the attention mechanism layer can focus on text paragraphs related to abnormal information, reducing the interference of noise information and improving the targeting and accuracy of the model.

[0089] The preset prompt template defines the basic structure and requirements of the reported content. The fusion layer can then be used to concatenate the first text vectors corresponding to the preset prompt template and the second text vectors corresponding to the text paragraphs in the vector space to generate several concatenated third text vectors. The first text vectors corresponding to the preset prompt template include: unit text vector, time text vector, method text vector, and template text vector. The second text vectors corresponding to the text paragraphs include: reporting text vector, reporting time text vector, reporting method text vector, and reporting template text vector.

[0090] Based on each concatenated third text vector, generate the reporting content corresponding to the preset prompt template.

[0091] The output layer is used to output the generated reporting content in a standardized format for subsequent processing or display.

[0092] The fusion layer of this invention combines the structural requirements of the preset prompt template with the specific content of the text paragraphs, generating a concatenated third text vector through vector concatenation. These vectors contain both the basic structure and requirements of the preset prompt template and the text content most relevant to the anomaly information extracted from the target procedure document. Therefore, the fusion layer ensures that the generated reporting content not only conforms to the specifications but also accurately reflects the anomaly information entered by the user.

[0093] Therefore, by processing the target procedure document through a preset document processing model, content related to power grid anomalies can be automatically extracted, thereby generating corresponding anomaly reporting information and improving the efficiency and accuracy of anomaly reporting.

[0094] For step S4, in a preferred embodiment, the abnormal text includes: the location of the abnormality, the time of the abnormality, the type of the abnormal device, and the abnormal status; the abnormal status is used to characterize the device status when the abnormality occurs; the reporting template includes: a time filling part, a unit filling part, a device type filling part, and an abnormal status filling part;

[0095] The step of generating abnormal reporting information corresponding to the power grid abnormal information based on each abnormal text and the reported content includes:

[0096] Fill the time of the anomaly, the reporting unit, the type of the abnormal device, and the state of the anomaly into the time, unit, device type, and anomaly situation fields of the reporting template respectively to generate a filled reporting template.

[0097] Based on the filled-in reporting template, the abnormal text of the abnormal location, the reporting time in the reporting content, and the reporting method in the reporting content, the abnormal reporting information corresponding to the power grid abnormal information is generated.

[0098] Specifically, the following is a concrete example illustrating how to generate abnormal reporting information corresponding to power grid abnormal information based on abnormal text and reporting content:

[0099] Example of abnormal text: Location of abnormality: XX Road Substation, XX District, XX City; Time of abnormality: 14:30 on May 15, 2023; Type of abnormal equipment: Transformer; Abnormal status: Overheating, temperature exceeds the set threshold;

[0100] Example of reported content (extracted from the target procedure document): Reporting unit: XX Power Company Operation and Maintenance Department; Reporting time: May 15, 2023, 14:35; Reporting method: Telephone report + system entry;

[0101] Reporting Template: Time field: [Time]; Unit field: [Unit]; Equipment type field: [Equipment Type]; Abnormal situation field: [Abnormal situation];

[0102] Indicatively, the template may also include other fixed text, such as "Dear Leader, the following power grid anomaly is reported:" etc.

[0103] When generating the pre-filled reporting template, the time of the anomaly occurrence, "May 15, 2023, 14:30," can be filled into the time field; the reporting unit, "XX Power Company Operation and Maintenance Department," can be filled into the unit field; the type of the abnormal equipment, "Transformer," can be filled into the equipment type field; and the abnormal status, "Overheating, temperature exceeds the set threshold," can be filled into the abnormal status field. The final pre-filled reporting template is as follows:

[0104] Dear Leader, I am reporting the following abnormal situation in the power grid:

[0105] Time: 14:30, May 15, 2023;

[0106] Unit: Operations and Maintenance Department of XX Power Company;

[0107] Equipment type: Transformer;

[0108] Abnormal situation: Overheating, the temperature exceeds the set threshold.

[0109] Furthermore, based on the aforementioned filled-in reporting template and other information, the process of generating the final reporting information is as follows:

[0110] Based on the already filled reporting template, add the location of the anomaly as "XX City XX District XX Road Substation"; add the reporting time extracted from the reporting content as "May 15, 2023, 14:35"; this is illustrative, although the time is close to the time of the anomaly, it is the actual reporting time; add the reporting method as "telephone report + system entry".

[0111] The final generated exception reporting information is as follows:

[0112] Dear Leader, I am reporting the following abnormal situation in the power grid:

[0113] Time: 14:30, May 15, 2023;

[0114] Location: Substation on XX Road, XX District, XX City;

[0115] Unit: Operations and Maintenance Department of XX Power Company;

[0116] Reporting time: 14:35 on May 15, 2023;

[0117] Reporting methods: telephone report + system entry;

[0118] Equipment type: Transformer;

[0119] Abnormal situation: Overheating, the temperature exceeds the set threshold.

[0120] Therefore, the filling method of this invention is based on the direct extraction of abnormal text and reported content, reducing the possibility of human input errors. At the same time, the accuracy of the information is ensured through explicit filling correspondences (such as filling the time of the abnormality to the time-filled section). Furthermore, by using the reporting template extracted from the procedure document, the format and content of the abnormal reporting information are ensured to be consistent and standardized with the procedure document, reducing misunderstandings or delays caused by inconsistent formats or incomplete information.

[0121] By automating the filling method and generation process, the preparation time for reporting information can be significantly shortened. At the same time, the standardized format simplifies the recipient's processing flow, improves work efficiency, and ensures that subsequent anomaly reports based on this information are accurately delivered to the appropriate units. This facilitates information sharing and collaborative work between different departments and teams, thereby enhancing the overall ability to respond to power grid anomalies.

[0122] In a preferred embodiment, the present invention may employ RAG technology, namely retrieval enhancement generation technology. By efficiently indexing and vectorizing the procedure documents, the applicable procedure documents can be quickly located when the user inputs an abnormal situation. By using RAG technology to retrieve relevant procedure documents in real time, abnormal events can be automatically matched with corresponding processing specifications, ensuring that information is timely and accurate.

[0123] In a preferred embodiment, the document processing model of the present invention is a Large Language Model (LLM). By combining the natural language processing capabilities of LLM, the present invention can deeply understand the context of the abnormal information input by the user and accurately extract the key information from the retrieved target procedure document, making the generation of reporting information more targeted and accurate, and effectively ensuring the consistency and completeness of abnormal reporting information.

[0124] Therefore, the embodiments of the present invention combine RAG technology with LLM to achieve rapid information retrieval and intelligent analysis, forming an integrated solution and optimizing the information processing flow.

[0125] like Figure 2 As shown, based on the embodiments of the above-mentioned methods for generating abnormal reporting information in various power distribution networks, the present invention provides corresponding device embodiments;

[0126] An embodiment of the present invention provides an anomaly reporting information generation device for a power distribution network, comprising: an anomaly information acquisition module, a similarity calculation module, a reporting content extraction module, and an anomaly reporting information generation module;

[0127] The abnormal information acquisition module is used to acquire power grid abnormal information input by the user; wherein, the power grid abnormal information includes: a number of abnormal texts used to describe abnormal phenomena;

[0128] The similarity calculation module is used to convert the power grid anomaly information into a semantic vector list, and calculate the similarity between each procedure document and the semantic vector list based on each semantic vector in the semantic vector list and the index vector of each procedure document; wherein, each semantic vector in the semantic vector list corresponds to an anomaly text; the procedure document is used to represent the reporting rules corresponding to the handling of power grid anomalies.

[0129] The reporting content extraction module is used to select the procedure document with the highest similarity as the target procedure document, and then extract the corresponding reporting content in the target procedure document; wherein, the reporting content includes: the reporting unit, the reporting time, and the reporting method;

[0130] The abnormal reporting information generation module is used to generate abnormal reporting information corresponding to the power grid abnormal information based on each abnormal text and the reporting content.

[0131] It should be noted that the device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0132] Those skilled in the art will clearly understand that, for convenience and simplicity, the specific working process of the device described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0133] Based on the above embodiments of various methods for generating abnormal reporting information in power distribution networks, the present invention provides corresponding embodiments for terminal equipment.

[0134] One embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a method for generating abnormal reporting information of a power distribution network as described in any embodiment of the present invention.

[0135] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0136] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0137] The memory can be used to store the computer program. The processor implements various functions of the terminal device by running or executing the computer program stored in the memory and calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created based on the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0138] Based on the above embodiments of various methods for generating abnormal reporting information in power distribution networks, the present invention provides corresponding embodiments for storage media.

[0139] One embodiment of the present invention provides a storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute a method for generating abnormal reporting information of a power distribution network as described in any embodiment of the present invention.

[0140] The storage medium is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When executed by a processor, the computer program can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0141] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for generating anomaly reporting information in a power distribution network, characterized in that, include: Obtain power grid anomaly information input by the user; wherein, the power grid anomaly information includes: several anomaly texts used to describe the anomaly phenomenon; The power grid anomaly information is converted into a semantic vector list. Based on the semantic vectors in the semantic vector list and the index vector of each procedure document, the similarity between each procedure document and the semantic vector list is calculated. Each semantic vector in the semantic vector list corresponds to an anomaly text. The procedure document is used to represent the reporting rules corresponding to the handling of power grid anomalies. The procedure document with the highest similarity is used as the target procedure document, and a preset prompt template for guiding the model to output preset content is obtained; wherein, the preset prompt template includes: reporting unit, reporting time, reporting method and reporting template; Each abnormal text in the power grid abnormal information is segmented into words, and a word segmentation list corresponding to the power grid abnormal information is generated based on each segmented word. The preset prompt template, target procedure document, and word segmentation list are input into a preset document processing model. The document processing model extracts word features of each word in the word segmentation list and captures contextual information of each text paragraph in the target procedure document. Based on the contextual information of each text paragraph, corresponding text features are generated. Based on each word feature and each text feature, the text paragraph with the highest relevance to the word segmentation list is extracted from the target procedure document. Based on the preset prompt template and the text paragraph, the reporting content corresponding to the preset prompt template is generated. Based on each abnormal text and the reported content, generate abnormal reporting information corresponding to the power grid abnormal information; The generation of the preset document processing model includes: Training samples are generated based on the word segmentation list samples corresponding to the procedure document samples, the power grid anomaly information samples, and the preset prompt template samples; wherein, the preset prompt template samples include: reporting unit, reporting time, reporting method, and reporting template; The document processing model to be trained is iteratively trained using each training sample and the actual reported content corresponding to each training sample as input, and the predicted reported content of each training sample as output, until the model converges, thus generating the preset document processing model.

2. The method for generating abnormal reporting information in a power distribution network as described in claim 1, characterized in that, The step of calculating the similarity between each procedure document and the semantic vector list based on each semantic vector in the semantic vector list and the index vector of each procedure document includes: For each procedure document, the similarity between each semantic vector in the semantic vector list and the index vector of the procedure document is calculated to obtain the similarity between each semantic vector and the procedure document. For each procedure document, the average similarity between the procedure document and the semantic vector list is obtained based on the similarity between each semantic vector in the semantic vector list and the procedure document, as well as the total number of semantic vectors in the semantic vector list; the average similarity is used as the similarity between the procedure document and the semantic vector list.

3. The method for generating abnormal reporting information in a power distribution network as described in claim 2, characterized in that, During each iteration of training, a training sample is input into the document processing model so that the document processing model can extract the text paragraph with the highest relevance to the word segmentation list sample from the procedure document sample based on the word features and text features in the training sample, and generate the reporting content prediction result corresponding to the preset prompt template sample based on the preset prompt template and the text paragraph. The predicted results of the reported content are compared with the actual reported content, and the network parameters of the document processing model are adjusted based on the comparison results.

4. The method for generating abnormal reporting information in a power distribution network as described in claim 3, characterized in that, The document processing model includes: an attention mechanism layer; The step of extracting the text paragraphs with the highest relevance to the word segmentation list from the target procedure document based on each word feature and each text feature includes: The attention mechanism layer is used to calculate the cosine similarity between each word feature and each text feature. For each text feature, based on the activation function, the cosine similarity between the text feature and each word feature is converted into a probability distribution, and attention weights are assigned to the text features according to each probability distribution; wherein, the attention weights are used to characterize the degree of correlation between the text features and the word segmentation list; Based on each attention weight, the text paragraph corresponding to the text feature with the highest attention weight is determined.

5. The method for generating abnormal reporting information in a distribution network as described in claim 4, characterized in that, The document processing model includes: a fusion layer; The step of generating reporting content corresponding to the preset prompt template based on the preset prompt template and the text paragraph includes: The fusion layer is used to concatenate the first text vectors corresponding to the preset prompt template and the second text vectors corresponding to the text paragraphs in the vector space to generate several concatenated third text vectors; wherein, the first text vectors corresponding to the preset prompt template include: unit text vector, time text vector, method text vector, and template text vector; the second text vectors corresponding to the text paragraphs include: reporting text vector, reporting time text vector, reporting method text vector, and reporting template text vector; Based on each concatenated third text vector, generate the reporting content corresponding to the preset prompt template.

6. The method for generating abnormal reporting information in a distribution network as described in claim 2, characterized in that, The abnormal text includes: the location of the abnormality, the time of the abnormality, the type of the abnormal device, and the abnormal status; the abnormal status is used to characterize the device status when the abnormal situation occurs; the reporting template includes: time filling part, unit filling part, device type filling part, and abnormal situation filling part; The step of generating abnormal reporting information corresponding to the power grid abnormal information based on each abnormal text and the reported content includes: Fill the time of the anomaly, the reporting unit, the type of the abnormal device, and the state of the anomaly into the time, unit, device type, and anomaly situation fields of the reporting template respectively to generate a filled reporting template. Based on the filled-in reporting template, the abnormal text of the abnormal location, the reporting time in the reporting content, and the reporting method in the reporting content, the abnormal reporting information corresponding to the power grid abnormal information is generated.

7. A device for generating abnormal reporting information in a power distribution network, characterized in that, include: The module includes an anomaly information acquisition module, a similarity calculation module, a report content extraction module, and an anomaly report information generation module. The abnormal information acquisition module is used to acquire power grid abnormal information input by the user; wherein, the power grid abnormal information includes: a number of abnormal texts used to describe abnormal phenomena; The similarity calculation module is used to convert the power grid anomaly information into a semantic vector list, and calculate the similarity between each procedure document and the semantic vector list based on each semantic vector in the semantic vector list and the index vector of each procedure document; wherein, each semantic vector in the semantic vector list corresponds to an anomaly text; the procedure document is used to represent the reporting rules corresponding to the handling of power grid anomalies. The reporting content extraction module is used to select the most similar procedure document as the target procedure document and obtain a preset prompt template for guiding the model to output preset content. The preset prompt template includes: reporting unit, reporting time, reporting method, and reporting template. It performs word segmentation on each abnormal text in the power grid anomaly information, generating a word segmentation list corresponding to the power grid anomaly information based on each segmented word. The preset prompt template, the target procedure document, and the word segmentation list are input into a preset document processing model, enabling the model to extract word features from each word in the word segmentation list and capture the context information of each text paragraph in the target procedure document, generating corresponding text features based on the context information of each text paragraph. Based on each word feature and each text feature, the module extracts the text paragraph with the highest relevance to the word segmentation list from the target procedure document, and generates reporting content corresponding to the preset prompt template based on the preset prompt template and the text paragraph. The generation of the preset document processing model includes: generating training samples based on the word segmentation list samples corresponding to the procedure document samples and the power grid anomaly information samples, as well as the preset prompt template samples; wherein the preset prompt template samples include: reporting unit, reporting time, reporting method, and reporting template; taking each training sample and the actual reported content corresponding to each training sample as input, and the predicted reported content of each training sample as output, iteratively training the document processing model to be trained until the model converges, thereby generating the preset document processing model; The abnormal reporting information generation module is used to generate abnormal reporting information corresponding to the power grid abnormal information based on each abnormal text and the reporting content.

8. A terminal device, characterized in that, The device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement a method for generating abnormal reporting information for a power distribution network as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium includes a stored computer program, wherein, when the computer program is running, it controls the device where the storage medium is located to execute a method for generating abnormal reporting information for a power distribution network as described in any one of claims 1 to 6.

Images