Power information triplet fault priority detection method, system and device

Abstract

The present invention belongs to the field of power information text data processing, and provides a power information triplet fault priority detection method, system and device. Based on the power information positioning matrix, a power information positioning center matrix is ​​generated, and the distribution positioning and distance positioning between each power information positioning matrix and the center matrix are calculated respectively; the power information dispersion distance of each power information triplet is generated according to the comparison of distribution positioning and distance positioning, and the power information triplet is sorted based on the power information dispersion distance to obtain the power information triplet fault priority sequence, and the fault problem is processed in the order of the priority sequence. The priority detection and processing of fundamental faults in the power system is realized, and the efficiency and accuracy of power grid fault troubleshooting are significantly improved.

Description

Technical Field

[0001] The present invention belongs to the field of power information text data processing, and in particular relates to a power information triplet fault priority detection method, system and equipment. Background Art

[0002] In the power system, the monitoring and troubleshooting of power equipment are key links to ensure the safe and stable operation of the power grid. However, with the expansion of the scale of the power grid and the increase in the complexity of equipment, the types of faults and the scope of impact are becoming more and more diverse. Power servers in various regions generate a large amount of fault record text, which contains information such as equipment location, fault type, fault description, etc. These text data need to be efficiently processed and analyzed to quickly identify potential faults in the power grid. In actual power networks, power information texts collected from multiple servers often represent different fault problems in the power grid structure, but it is difficult to determine the key faults to be prioritized among these fault problems. Sometimes, if the root fault problem is not solved in time, prioritizing surface faults will lead to repeated faults.

[0003] Generative information extraction models are widely used to process information of power equipment, mainly referring to the automatic extraction of triple data (i.e., SPO triples) from a large amount of text data of power equipment. The key goal of this model is to convert the original unstructured text information, such as operation records, defect reports, alarm logs, etc., into structured data to provide support for power equipment status analysis, prediction and management. These data mainly come from multiple information systems of the power system, such as: equipment ledgers, which contain basic information of the equipment (such as model, location, manufacturer, etc.); monitoring logs, such as equipment monitoring data and alarm information automatically generated by the system; defect and maintenance records, which contain manual or automatic data such as maintenance history, defect reports, and test results; text information, which contains a large number of detailed descriptions of equipment status, abnormal conditions, and maintenance operations, such as: "The pressure of the hydraulic mechanism of the high 2201 switch of the 110kV experimental substation #1 main transformer is low and has not reached the locking value" and "The temperature of the A phase winding of the 220kV main transformer exceeds the standard, and the alarm temperature exceeds the threshold". Triple data is a structured data expression method, which usually represents entities and their relationships in power equipment information in this form. This structured information is easier for computers to process and analyze. For example, the triples extracted from the text "110kV experimental substation #1 main transformer high 2201 switch hydraulic mechanism pressure is low and has not reached the lock value" can be: (main transformer high 2201 switch, abnormal situation, low pressure), (hydraulic mechanism, state, has not reached the lock value), etc. Compared with traditional information extraction methods, generative models deal with unstructured and diverse texts. The operation and maintenance records of power equipment are complex and diverse, and traditional rules and template methods are difficult to adapt efficiently. Generative models can identify key entities and relationships in sentences through the context understanding ability of pre-trained language models without relying on fixed templates. Generative models can convert implicit information in long texts into usable structured data through the structure of SPO triples, providing support for knowledge graphs, predictive analysis, and fault diagnosis. Generative models can migrate between multiple tasks and adapt to different types of power equipment texts (such as defect records and alarm logs) through simple fine-tuning, making them more adaptable in practical applications.

[0004] Traditional power system fault detection methods mostly use an alarm system based on preset thresholds. When a certain operating parameter of the power equipment exceeds the set range, the system triggers an alarm, and the maintenance personnel check the equipment according to the alarm log. For example, in the power equipment fault analysis method provided in the patent document with publication number CN117235254A, although it is possible to perform fault analysis matching based on the text clustering results, it is difficult to consider the correlation between different faults at the same time, especially in multi-level linkage faults. If the fundamental fault is not resolved, it may cause other related equipment to frequently have surface faults. It is also difficult to distinguish between fundamental faults and minor faults using ordinary language models. Existing systems usually troubleshoot faults in the order in which the alarms are generated, without clear priority sorting. This will result in some minor surface faults being handled first, while key faults will not be resolved in time due to insufficient priority. For example, a patent document with publication number CN101917001B provides a fault sorting method for online static safety analysis of power systems. Although it can overcome the disadvantage of low calculation accuracy when relying solely on traditional behavioral indicators for fault sorting, it ignores the fact that power fault information usually exists in the form of text records, and traditional methods lack efficient text information extraction and analysis methods, making it difficult to comprehensively and accurately extract the core elements of the fault, such as equipment location, fault type, fault description, etc., and thus cannot provide refined fault troubleshooting guidance. Summary of the invention

[0005] The purpose of the present invention is to propose a method, system and device for prioritizing detection of power information triplet faults, so as to solve one or more technical problems existing in the prior art and at least provide a beneficial choice or create conditions.

[0006] In order to achieve the above object, according to one aspect of the present invention, a method for fault priority detection of power information triplet is provided, the method comprising the following steps:

[0007] Collecting power information text, extracting multiple power information triplets including equipment location, fault type and fault description text information from the text, vectorizing and embedding the elements in each triplet respectively, and generating vector representations of each element;

[0008] Calculating the corresponding power information ternary covariance matrix for the element vector of each power information triplet, and generating a power information positioning matrix based on the eigenvalues ​​and eigenvectors of the power information ternary covariance matrix;

[0009] Generate a power information positioning center matrix based on the power information positioning matrix, and respectively calculate the distribution positioning and distance positioning between each power information positioning matrix and the power information positioning center matrix;

[0010] The power information dispersion distance of each power information triplet is generated according to the comparison between the distribution positioning and the distance positioning, and the power information triplet is sorted based on the power information dispersion distance to obtain the power information triplet fault priority sequence, and the fault problem is processed in the order of the priority sequence.

[0011] Furthermore, the RoUIE model is used as a triple extraction model for power information text, rotation position encoding and distribution focus loss are introduced into the RoUIE model, and the embedding vector of the RoUIE model is used to vectorize and embed the elements in each triple.

[0012] Furthermore, the Roformer v2 model is used as an encoder as a triple extraction model for power information text, the Roformer v2 model introduces a loss function combining fuzzy span loss and Wasserstein distance, and the embedding vector of the Roformer v2 model is used to vectorize and embed the elements in each triplet respectively.

[0013] Furthermore, in the power information ternary covariance matrix, each matrix unit is composed of the vector covariance of two elements in the power information triplet.

[0014] Furthermore, the power information locating quantity is used instead of the power information locating matrix. The power information locating quantity is represented by an array generated by sorting the eigenvalues ​​of the power information ternary covariance matrix according to numerical values.

[0015] Furthermore, by performing cluster analysis on each power information positioning matrix, a number of cluster centers are obtained as a number of power information positioning center matrices;

[0016] The power information positioning matrix is ​​a matrix representation generated by sorting the eigenvectors corresponding to the eigenvalues ​​of the power information ternary covariance matrix according to the numerical values ​​of the eigenvalues.

[0017] Furthermore, the power information triplet fault priority sequence is obtained by sorting each power information triplet from large to small according to the power information dispersion distance, so as to give priority to processing the fault problems represented by each power information triplet in the power information triplet fault priority sequence.

[0018] Furthermore, the distribution positioning is obtained by calculating the information entropy value of each power information positioning matrix relative to the power information positioning center matrix;

[0019] The distance positioning is obtained by calculating the distance of each power information positioning matrix relative to the power information positioning center matrix;

[0020] The power information dispersion distance is obtained by comparing the reverse growth trend between the distribution location value of each power information triplet and the corresponding distance location value.

[0021] The present invention also provides a power information triplet fault priority detection system, the power information triplet fault priority detection system comprises: a processor, a memory, and a computer program stored in the memory and executable on the processor, the processor implements the steps in the power information triplet fault priority detection method when executing the computer program, the power information triplet fault priority detection system can be run in computing devices such as desktop computers, laptop computers, PDAs, and cloud data centers, and the executable system may include, but is not limited to, a processor, a memory, and a server cluster, and the processor executes the computer program to run in the following system units:

[0022] An information extraction unit, used to collect power information text, extract multiple power information triplets containing equipment location, fault type and fault description text information from the text, vectorize and embed the elements in each triplet, and generate vector representations of each element;

[0023] A matrix operation unit, used for calculating the corresponding power information ternary covariance matrix for the element vector of each power information triplet, and generating a power information positioning matrix based on the eigenvalues ​​and eigenvectors of the power information ternary covariance matrix;

[0024] A positioning calculation unit, used to generate a power information positioning center matrix based on the power information positioning matrix, and respectively calculate the distribution positioning and distance positioning between each power information positioning matrix and the power information positioning center matrix;

[0025] A priority detection unit is used to generate a power information dispersion distance of each power information triplet based on the comparison between the distribution positioning and the distance positioning, and to sort the power information triplet based on the power information dispersion distance to obtain a power information triplet fault priority sequence, and to process the fault problem in the order of the priority sequence.

[0026] Correspondingly, the present invention also provides an electronic device, a readable storage medium and a computer program product:

[0027] An electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the power information triplet fault priority detection method and the methods of each step therein.

[0028] A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to execute the power information triplet fault priority detection method and the steps therein.

[0029] A computer program product comprises a computer program, wherein when the computer program is executed by a processor, the method for detecting fault priority of a power information triplet and each step thereof are implemented.

[0030] The beneficial effects of the present invention are as follows: the present invention provides a method, system and device for fault priority detection of power information triples, calculates the corresponding power information ternary covariance matrix for the element vector of each power information triple, and generates a power information positioning matrix based on the eigenvalues ​​and eigenvectors of the power information ternary covariance matrix; generates a power information positioning center matrix based on the power information positioning matrix, and respectively calculates the distribution positioning and distance positioning between each power information positioning matrix and the center matrix; generates the power information dispersion distance of each power information triple based on the comparison of distribution positioning and distance positioning, and sorts the power information triples based on the power information dispersion distance to obtain the power information triple fault priority sequence, and handles the fault problem in the order of the priority sequence. The priority detection and processing of fundamental faults in the power system is realized, which significantly improves the efficiency and accuracy of power grid fault troubleshooting.

[0031] The present invention uses models such as RoUIE or Roformer v2 to extract core information such as equipment location, fault type, fault description, etc. in the power information text as triples, and vectorizes each element to form an embedded representation. Unlike traditional fault detection based on keywords or a single threshold, this method performs semantic analysis of the text through a deep learning model to achieve more comprehensive and accurate information extraction. By using deep learning models such as RoUIE or Roformer v2 to extract power information triplets, and vectorizing and embedding the triplet elements, the system can accurately capture the fault elements and their interrelationships in the text information. This process retains the semantic information and relative position relationship of each element, making the fault information representation more comprehensive.

[0032] The present invention performs covariance calculation on the vector of each power information triplet and constructs a covariance matrix to capture the mutual correlation of the elements in the triplet. The construction of the covariance matrix effectively makes up for the shortcomings of the existing methods in fault correlation analysis, and helps to identify the correlation between fundamental faults and surface faults. The quantitative expression of fault information is achieved through covariance matrix and eigenvalue analysis. Combined with the priority sequence generated by distribution positioning, distance positioning and distribution distance calculation, the system can identify and prioritize fundamental faults. Compared with the time sorting method of the prior art, this method is more in line with the actual troubleshooting needs of power system faults.

[0033] The present invention further analyzes the covariance matrix through eigenvalues ​​and eigenvectors to generate a power information positioning matrix. By clustering the positioning matrix, a power information positioning center matrix is ​​generated, and the distribution distance of each fault triplet is calculated by combining distribution positioning and distance positioning. Priority sorting based on distribution distance helps the system to put fundamental faults in the forefront to ensure priority processing, avoiding resource waste and processing delays caused by disorderly troubleshooting in existing methods. The technical solution of the present invention enables the system to automatically extract, analyze, sort and process fundamental faults in a variety of fault information, reducing errors in manual judgment and improving the efficiency and accuracy of fault troubleshooting. This solution reduces the risk of large-scale failures in the power grid while reducing system maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The above and other features of the present invention will become more obvious by describing in detail the embodiments shown in the accompanying drawings. The same reference numerals in the accompanying drawings of the present invention represent the same or similar elements. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other accompanying drawings can be obtained based on these accompanying drawings without creative work. In the accompanying drawings:

[0035] Figure 1 Shown is a flow chart of a power information triplet fault priority detection method;

[0036] Figure 2 Shown is the system structure diagram of the power information triplet fault priority detection system. DETAILED DESCRIPTION

[0037] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0038] In the description of the present invention, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed" etc. are understood as not including the number itself, and "above", "below", "within" etc. are understood as including the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0039] like Figure 1 The flowchart of the power information triplet fault priority detection method according to the present invention is shown below. Figure 1 To illustrate the power information triplet fault priority detection method, system and device according to the implementation mode of the present invention.

[0040] The present invention proposes a method for fault priority detection of a power information triplet, which specifically comprises the following steps:

[0041] Collecting power information text, extracting multiple power information triplets including equipment location, fault type and fault description text information from the text, vectorizing and embedding the elements in each triplet respectively, and generating vector representations of each element;

[0042] Calculating the corresponding power information ternary covariance matrix for the element vector of each power information triplet, and generating a power information positioning matrix based on the eigenvalues ​​and eigenvectors of the power information ternary covariance matrix;

[0043] Generate a power information positioning center matrix based on the power information positioning matrix, and respectively calculate the distribution positioning and distance positioning between each power information positioning matrix and the center matrix;

[0044] The power information dispersion distance of each power information triplet is generated according to the comparison between the distribution positioning and the distance positioning, and the power information triplet is sorted based on the power information dispersion distance to obtain the power information triplet fault priority sequence, and the fault problem is processed in the order of the priority sequence.

[0045] In some embodiments, traditional power fault detection methods are often based on preset fault types or manual troubleshooting, and it is difficult to systematically determine the correlation and priority of different faults. The technical solution provided by the present invention forms a systematic priority sorting process through embedding and covariance matrix construction, which has a higher degree of automation. This technical solution realizes the priority sorting of fault problems, and effectively prioritizes fundamental faults by calculating the power information dispersion distance. Compared with the traditional troubleshooting method based on alarm sequence, this method is more efficient and avoids the situation of repairing surface problems without solving the fundamental problems, thereby improving the accuracy of power grid fault troubleshooting. The scheme of the present invention determines the correlation between triplets based on covariance matrix analysis and clustering, and uses eigenvalues, eigenvectors and power information dispersion distances to determine the priority of faults. The covariance matrix can quantify the correlation between elements of different triplets, and the positioning matrix formed by eigenvalue sorting combined with the power information dispersion distance of the cluster center helps to identify fundamental fault problems with higher abnormalities.

[0046] Furthermore, the RoUIE model is used as the triple extraction model of the power information text, the rotation position encoding and distribution focus loss are introduced into the RoUIE model, and the embedding vector of the RoUIE model is used to vectorize and embed the elements in each triple, specifically:

[0047] In some optional embodiments, the RoUIE model is introduced as a triple extraction model, and the rotation position encoding and distribution focus loss are used to improve the quality of embedding, and the embedding vector of the RoUIE model is used for the vectorized embedding of triples. The RoUIE model combines rotation position encoding and distribution focus loss, making the model more stable and accurate when processing triple extraction in power text. When extracting triples, traditional models are difficult to effectively handle the relative position changes of elements in the text, resulting in reduced embedding quality. By using the RoUIE model and rotation position encoding, the extracted triple embedding vector more accurately retains the relational structure in the power information text, greatly improves the model's tolerance to position changes in the triple, and ensures that the embedding vector can more truly reflect the relative relationship between the elements. Rotation position encoding has the advantage of being insensitive to position changes in the text sequence, so that the embedding vector can capture the relative position relationship between elements in the power information triple. Distribution focus loss further optimizes the accuracy and distribution rationality of embedding by focusing on the distribution of abnormal data points.

[0048] Furthermore, the Roformer v2 model is used as an encoder as a triple extraction model for power information text, the Roformer v2 model introduces a loss function combining fuzzy span loss and Wasserstein distance, and the embedding vector of the Roformer v2 model is used to vectorize and embed the elements in each triplet respectively.

[0049] In other optional embodiments, the Roformer v2 model can be introduced as an encoder, and the loss functions of fuzzy span loss and Wasserstein distance can be combined to improve the vectorized embedding quality of triples. Roformer v2 processes complex power texts through fuzzy span loss and Wasserstein distance, so that the model can better extract fuzzy boundaries in triples. Compared with traditional vectorization methods, it has more advantages in processing noisy data and uncertainty boundaries. This method can more accurately process fuzzy boundaries in power information texts and improve the stability of vectorized embedding. The introduction of Wasserstein distance can ensure that the model can still maintain robustness when abnormal data points appear, providing support for the accuracy of subsequent fault priority sorting. Fuzzy span loss can still accurately extract features when the boundaries are fuzzy, and Wasserstein distance is used to measure the overall similarity between vectors, which has the characteristic of being insensitive to outliers. This combination effectively improves the embedding accuracy of complex texts.

[0050] Furthermore, in the power information ternary covariance matrix, each matrix unit is composed of the vector covariance of two elements in the power information triplet:

[0051] In some embodiments, in the power information triplet to be detected, each power information triplet has three elements, where the elements are strings of text representing power failure problems. For each power information triplet, the covariance between the vectors of the three elements in each power information triplet is calculated, and the covariance matrix composed of the covariance between the vectors of the elements is used as the power information ternary covariance matrix corresponding to the power information triplet.

[0052] In the power information ternary covariance matrix, each matrix unit is composed of the vector covariance of two elements to quantify the relationship between the elements in the triple. Through the covariance matrix, the traditional method cannot accurately quantify the complex relationship between the elements in the triple, while this scheme quantifies the relationship between the elements of the triple into a mathematical matrix through the covariance matrix, which provides a basis for subsequent feature analysis. The construction of the covariance matrix enables the correlation between the elements in the triple to be accurately expressed, which facilitates the use of eigenvalues ​​and eigenvectors to analyze these correlations, making the final priority sorting more reasonable. The covariance matrix can capture the correlation between vectorized embeddings and form quantitative features, so that the subsequent analysis of eigenvalues ​​and eigenvectors can reveal the potential correlation patterns between the elements of the triple.

[0053] Furthermore, the power information locating quantity is used instead of the power information locating matrix. The power information locating quantity is represented by an array generated by sorting the eigenvalues ​​of the power information ternary covariance matrix according to numerical values.

[0054] In some embodiments, the power information location quantity can be used as a substitute for the power information location matrix, and the location quantity can be represented by an array generated by eigenvalue sorting. If it is found that the calculation amount of selecting matrix operations under the current hardware conditions will be large and may not be easy to use directly for sorting analysis, then the power information location quantity obtained by eigenvalue sorting in this scheme can be selected to effectively reduce the calculation complexity while retaining the main characteristic information of the matrix. Eigenvalues ​​and eigenvectors can be pre-processed by unitization and normalization. The power information location quantity is represented by an array generated by eigenvalue sorting, which greatly simplifies the calculation while retaining the main characteristic information, and provides an efficient data processing method for subsequent clustering and sorting. Eigenvalue sorting can retain the main features in the covariance matrix, and using an array instead of a matrix to represent it simplifies the calculation process, and can be quickly applied to clustering and priority sorting.

[0055] Furthermore, by performing cluster analysis on each power information positioning matrix, a number of cluster centers are obtained as a number of power information positioning center matrices.

[0056] The power information positioning matrix is ​​a matrix representation generated by sorting the eigenvectors corresponding to the eigenvalues ​​of the power information ternary covariance matrix according to the numerical values ​​of the eigenvalues.

[0057] In some embodiments, several power information positioning center matrices can be generated by cluster analysis of the power information positioning matrix to identify the fault types and similarities in the power system. For the power information ternary covariance matrix corresponding to each power information triple, the eigenvalues ​​of the power information ternary covariance matrix and the eigenvectors corresponding to each eigenvalue are calculated respectively. The eigenvalues ​​of the power information ternary covariance matrix are sorted by numerical value, and the array formed is used as the power information positioning amount corresponding to the power information triple. The eigenvectors corresponding to the eigenvalues ​​of the power information ternary covariance matrix are sorted in order of their eigenvalue numerical values, and the matrix formed is used as the power information positioning matrix corresponding to the power information triple. It is difficult to find diversified fault modes through cluster analysis using traditional methods, and this scheme generates a central matrix through the positioning matrix, so that the system can automatically identify different fault types and similar modes. Through the central matrix, the system can realize the rapid identification of different types of faults, and effectively distinguish the fault categories when generating the priority sequence, and improve the accuracy of fault sorting. The central matrix clusters a large amount of similar data together, which helps to form a holistic understanding of the failure mode and avoids the computational burden of directly sorting the triplets individually.

[0058] Furthermore, the power information triplet fault priority sequence is obtained by sorting the power information dispersion distance from large to small, so as to give priority to processing the fault problems in the fault sequence.

[0059] In some embodiments, each power information triplet is sorted from large to small according to its corresponding power information dispersion distance to obtain a power information triplet fault sequence; and the fault problems corresponding to each power information triplet are processed according to the order of the power information triplet fault sequence.

[0060] Although a lot of power information texts can be extracted from the collected power information texts from various servers, each of which represents a fault problem in the power grid structure, it is difficult to know which fault problem to start troubleshooting from among the multiple power information texts. Sometimes, when a fundamental fault problem has not been solved, if you solve the superficial fault problems first, even if the superficial fault problems are repaired, the problems in the power transmission network structure itself are still not solved. Therefore, we must first select the fundamental key fault problems represented by the multiple power information texts, prioritize the fault problems represented by the multiple power information texts, and then select the key fault problems for repair, so that we can better use the power information triples for key troubleshooting and improve the efficiency and accuracy of troubleshooting and maintenance.

[0061] By sorting the power information dispersion distance from large to small, a power information triplet fault priority sequence is generated to give priority to fundamental fault problems. Compared with the existing method that cannot systematically sort fault priorities, this scheme generates a priority sequence by sorting the power information dispersion distance, which can more reasonably identify and prioritize fundamental faults, thereby avoiding repeated faults caused by unresolved fundamental problems. The fault priority sequence ensures that fundamental problems are given priority, prevents low-priority problems from covering up fundamental faults, and improves the maintenance efficiency and troubleshooting effect of the power system. The size of the power information dispersion distance represents the deviation of the triplet relative to the cluster center. The larger the power information dispersion distance, the more likely it is a core problem. Therefore, the sorting priority sequence ensures the priority treatment of fundamental problems.

[0062] Furthermore, the distribution positioning is obtained by calculating the information entropy value of each power information positioning matrix relative to the power information positioning central matrix, the distance positioning is obtained by calculating the distance of each power information positioning matrix relative to the power information positioning central matrix, and the power information dispersion distance is obtained by the inverse growth comparison trend between the distribution positioning value of each power information triplet and its corresponding distance positioning value.

[0063] In some embodiments, the cross entropy of the power information positioning matrix corresponding to each power information triplet relative to the power information positioning center matrix can be calculated as the power information distribution positioning corresponding to each power information triplet. The Euclidean distance of the power information positioning matrix corresponding to each power information triplet relative to the power information positioning center matrix can be calculated as the power information distance positioning corresponding to each power information triplet. The value of the power information distribution positioning corresponding to each power information triplet can be divided by the value of the power information distance positioning corresponding to it. When calculating, both can be first unitized, normalized, etc., as the power information dispersion distance corresponding to the power information triplet.

[0064] The distribution location is obtained through the information entropy value, the distance location is obtained through the distance, and the power information dispersion distance is generated through the comparative trend of distribution location and distance location. It is difficult for traditional systems to effectively introduce information entropy and distance calculation in fault location. This solution quantifies the abnormality of triples through information entropy and distance location, accurately measures distribution and geometric differences, and improves the accuracy of fault priority sorting. The power information dispersion distance combines distribution location and distance location, allowing the system to make an accurate distinction between fundamental faults and surface faults, thereby improving the detection and processing effects. Information entropy represents the degree of chaos in the distribution. Combined with the Euclidean distance, a comprehensive power information dispersion distance can be generated. Through this composite indicator, the system can identify the most influential fundamental faults and prioritize them.

[0065] For the power grid monitoring center to be tested, the computer system automatically collects real-time fault logs from the servers of the regional substations, power stations, and transmission lines to form multiple power information text records. These records mainly include equipment failures, abnormal alarms, and operation records. The system extracts and vectorizes each record, generates multiple power information triplets, such as text data of operation records such as equipment location, fault type, and fault details, and calculates the power information dispersion distance of each triplet to determine the priority of each fault problem.

[0066] During the routine detection process, the system collected the following five power information triplets representing different faults:

[0067] T1: The location is 110kV substation A, the fault type is voltage abnormality, and the description is "the voltage rises to the overload threshold."

[0068] T2: The location is 220kV main transformer, the fault type is temperature abnormality, described as "temperature exceeds 95℃".

[0069] T3: The location is transmission line B, the fault type is line tripping, and the description is "line tripping, resulting in a short power supply interruption."

[0070] T4: The location is 500kV substation C, the fault type is equipment damage, and the description is "mechanical failure of the circuit breaker".

[0071] T5: The location is 330kV substation D, the fault type is current abnormality, and the description is "current fluctuation exceeds the normal range".

[0072] After vectorizing the elements of each power information triplet, the system calculates the distribution location value and distance location value for each triplet, where:

[0073] The distribution positioning value of T1 is 0.75, and the distance positioning value is 0.25, so the power information dispersion distance is 3.0.

[0074] The distribution positioning value of T2 is 0.85, and the distance positioning value is 0.20, so the power information dispersion distance is 4.25.

[0075] The distribution positioning value of T3 is 0.65, and the distance positioning value is 0.35, so the power information dispersion distance is 1.86.

[0076] The distribution positioning value of T4 is 1.0, and the distance positioning value is 0.15, so the power information dispersion distance is 6.67.

[0077] The distribution positioning value of T5 is 0.60, and the distance positioning value is 0.40, so the power information dispersion distance is 1.5.

[0078] The power information triplets are sorted in descending order according to the power information dispersion distance to generate a fault priority sequence. The sorting results are as follows:

[0079] The power information dispersion distance of T4 is 6.67, with the highest priority, indicating that the circuit breaker of 500 kV substation C has a mechanical failure.

[0080] The power information dispersion distance of T2 is 4.25, indicating that the temperature of the 220kV main transformer exceeds 95℃.

[0081] The power information dispersion distance of T1 is 3.0, indicating that the voltage of 110kV substation A has increased to the overload threshold.

[0082] The power information dispersion distance of T3 is 1.86, indicating that a line trip occurred on transmission line B, resulting in a short power outage.

[0083] The power information dispersion distance of T5 is 1.5, indicating that the current fluctuation of 330kV substation D exceeds the normal range.

[0084] According to the fault priority sequence, the system operation sends instruction information in sequence to handle these faults:

[0085] Processing T4 (equipment damage): T4 means that the circuit breaker of 500kV substation C has a mechanical failure, which is the highest priority. The system immediately dispatches emergency repair personnel to 500kV substation C to replace or repair the circuit breaker. If this type of fault is not handled, it will affect the main transmission network and cause a wider range of power outages.

[0086] Processing T2 (temperature abnormality): T2 shows that the temperature of the 220kV main transformer is abnormal, and the temperature exceeds 95°C. The system will remotely control and reduce the load and check whether the cooling system is working properly. If necessary, it will arrange to send corresponding command information to engineers or automatic robots to maintain the transformer to prevent equipment from burning or causing more serious power supply problems.

[0087] Processing T1 (voltage abnormality): T1 indicates that the voltage of 110kV substation A has abnormally increased to the overload threshold. The system operator reduces the load distribution of substation A in the background control and monitors whether there is any grid load imbalance to ensure that the voltage returns to the normal range.

[0088] Processing T3 (line tripping): T3 involves a short tripping of transmission line B, which causes power outages in a small area. The system remotely restarts line B and checks whether there are external factors, such as strong winds, tree interference, etc., that cause the tripping. If the tripping is frequent, the system arranges an inspection team to conduct on-site inspections.

[0089] Processing T5 (current abnormality): T5 indicates that the current fluctuation of 330kV substation D exceeds the normal range, but does not reach the emergency level. The system records the abnormal situation and sets real-time monitoring to observe the subsequent fluctuation. If the fluctuation continues, arrange subsequent maintenance.

[0090] By sorting fault priorities and locating fundamental faults, the system prioritized the handling of fundamental faults and equipment damage in 500kV substation C, thus avoiding large-scale interruptions in the main power grid. Subsequently, the system processed secondary priority issues such as temperature anomalies and voltage anomalies in turn to ensure the stability and safety of power grid operation.

[0091] The power information triplet fault priority detection system runs in any computing device such as a desktop computer, a laptop computer, a PDA or a cloud data center. The computing device includes: a processor, a memory and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps in the power information triplet fault priority detection method are implemented. The executable system may include, but is not limited to, a processor, a memory, and a server cluster.

[0092] The embodiment of the present invention provides a power information triplet fault priority detection system, such as Figure 2 As shown, the power information triplet fault priority detection system of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the above-mentioned power information triplet fault priority detection method embodiment when executing the computer program, and the processor executes the computer program to run in the following system units:

[0093] An information extraction unit, used to collect power information text, extract multiple power information triplets containing equipment location, fault type and fault description text information from the text, vectorize and embed the elements in each triplet, and generate vector representations of each element;

[0094] A matrix operation unit, used for calculating the corresponding power information ternary covariance matrix for the element vector of each power information triplet, and generating a power information positioning matrix based on the eigenvalues ​​and eigenvectors of the power information ternary covariance matrix;

[0095] A positioning calculation unit, used to generate a power information positioning center matrix based on the power information positioning matrix, and respectively calculate the distribution positioning and distance positioning between each power information positioning matrix and the power information positioning center matrix;

[0096] A priority detection unit is used to generate a power information dispersion distance of each power information triplet based on the comparison between the distribution positioning and the distance positioning, and to sort the power information triplet based on the power information dispersion distance to obtain a power information triplet fault priority sequence, and to process the fault problem in the order of the priority sequence.

[0097] Among them, in order to better unify the numerical linear relationship and probabilistic connection between physical quantities of different units of measurement, different physical quantities can be dimensionally processed.

[0098] Among them, preferably, all undefined variables in the present invention, if not clearly defined, can be manually set thresholds.

[0099] The power information triplet fault priority detection system can be run in computing devices such as desktop computers, laptops, PDAs, and cloud data centers. The power information triplet fault priority detection system includes, but is not limited to, a processor and a memory. Those skilled in the art can understand that the examples are merely examples of power information triplet fault priority detection methods, systems, and devices, and do not constitute limitations on power information triplet fault priority detection methods, systems, and devices, and may include more or fewer components than the examples, or a combination of certain components, or different components. For example, the power information triplet fault priority detection system may also include input and output devices, network access devices, buses, and the like.

[0100] The present invention also provides an electronic device, a readable storage medium and a computer program product:

[0101] An electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the power information triplet fault priority detection method and the methods of each step therein.

[0102] A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to execute the power information triplet fault priority detection method and the steps therein.

[0103] A computer program product comprises a computer program, wherein when the computer program is executed by a processor, the method for detecting fault priority of a power information triplet and each step thereof are implemented.

[0104] Wherein, electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0105] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0106] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0107] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, 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 foregoing.

[0108] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0109] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0110] A computer system may include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other.

[0111] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete component gate circuits or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the power information triplet fault priority detection system, and uses various interfaces and lines to connect the various sub-areas of the entire power information triplet fault priority detection system.

[0112] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the power information triplet fault priority detection method, system and device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0113] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in the present disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in the present invention can be achieved, and this document is not limited here.

[0114] The present invention provides a method, system and device for fault priority detection of power information triples, calculates the corresponding power information triple covariance matrix for the element vector of each power information triple, and generates a power information positioning matrix based on the eigenvalues ​​and eigenvectors of the power information triple covariance matrix; generates a power information positioning center matrix based on the power information positioning matrix, and respectively calculates the distribution positioning and distance positioning between each power information positioning matrix and the center matrix; generates the power information dispersion distance of each power information triple according to the comparison of distribution positioning and distance positioning, and sorts the power information triples based on the power information dispersion distance to obtain the power information triple fault priority sequence, and handles the fault problem in the order of the priority sequence. The method realizes the priority detection and processing of fundamental faults in the power system, and significantly improves the efficiency and accuracy of power grid fault troubleshooting.

[0115] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A power information triplet fault priority detection method, characterized in that: The method comprises the following steps: Collecting power information text, extracting multiple power information triplets including equipment location, fault type and fault description text information from the text, vectorizing and embedding the elements in each triplet respectively, and generating vector representations of each element; Calculating the corresponding power information ternary covariance matrix for the element vector of each power information triplet, and generating a power information positioning matrix based on the eigenvalues ​​and eigenvectors of the power information ternary covariance matrix; Generate a power information positioning center matrix based on the power information positioning matrix, and respectively calculate the distribution positioning and distance positioning between each power information positioning matrix and the power information positioning center matrix; The power information dispersion distance of each power information triplet is generated according to the comparison between the distribution positioning and the distance positioning, and the power information triplet is sorted based on the power information dispersion distance to obtain the power information triplet fault priority sequence, and the fault problem is processed in the order of the priority sequence.

2. The power information triplet fault priority detection method according to claim 1, characterized in that: in, The RoUIE model is used as the triple extraction model of power information text. Rotational position encoding and distribution focus loss are introduced into the RoUIE model, and the embedding vector of the RoUIE model is used to vectorize and embed the elements in each triple.

3. The power information triplet fault priority detection method according to claim 1, characterized in that: in, The Roformer model is used as an encoder for a triple extraction model of power information text. The Roformer model introduces a loss function combining fuzzy span loss and Wasserstein distance, and the embedding vector of the Roformer model is used to vectorize and embed the elements in each triple.

4. The power information triplet fault priority detection method according to claim 1, characterized in that: In the power information ternary covariance matrix, each matrix unit is composed of the vector covariance of two elements in the power information triplet.

5. The power information triplet fault priority detection method according to any one of claims 1 to 3, characterized in that: The power information locating quantity is used instead of the power information locating matrix. The power information locating quantity is represented by an array generated by sorting the eigenvalues ​​of the power information ternary covariance matrix according to numerical values.

6. The power information triplet fault priority detection method according to any one of claims 1 to 3, characterized in that: in, By performing cluster analysis on each power information location matrix, a number of cluster centers are obtained as a number of power information location center matrices; The power information positioning matrix is ​​a matrix representation generated by sorting the eigenvectors corresponding to the eigenvalues ​​of the power information ternary covariance matrix according to the numerical values ​​of the eigenvalues.

7. The method according to claim 6, characterized in that in, The power information triplet fault priority sequence is obtained by sorting the power information triplet from large to small according to the power information dispersion distance, so as to give priority to processing the fault problems represented by each power information triplet in the power information triplet fault priority sequence.

8. The power information triplet fault priority detection method according to claim 7, characterized in that: The distribution positioning is obtained by calculating the information entropy value of each power information positioning matrix relative to the power information positioning center matrix; The distance positioning is obtained by calculating the distance of each power information positioning matrix relative to the power information positioning center matrix; The power information dispersion distance is obtained by comparing the reverse growth trend between the distribution location value of each power information triplet and the corresponding distance location value.

9. The power information triplet fault priority detection system is characterized by: The power information triplet fault priority detection system runs in any computing device of a desktop computer, a laptop computer or a cloud data center, and the computing device includes: a processor, a memory and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, it implements the steps in the power information triplet fault priority detection method as described in any one of claims 1 to 4.

10. An electronic device, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 4.

Images