Method and device for determining familial defects of power equipment

Abstract

The invention provides a method and device for determining familial defects of power equipment, and relates to the technical field of operation and maintenance of the power equipment. The main purpose of the technical scheme is to establish an objective and unified judgment standard so as to more accurately judge whether the electrical equipment has familial defects or not. According to the main technical scheme, the method comprises the steps of obtaining historical familial defect data and corresponding target features based on power equipment of a target category; based on the target feature, traversing a preset knowledge graph model to determine the occurrence frequency of the target feature, the preset knowledge graph model being an association relationship model constructed based on the operation data of the target power equipment and used for representing the defect event of the power equipment; based on the historical familial defect data, using a feature analysis algorithm to determine a weight coefficient of a target feature; and based on the occurrence frequency and the weight coefficient, calculating a suspected familial defect rate to determine that the target power equipment has the familial defect.

Description

Technical Field

[0001] This application relates to the field of power equipment operation and maintenance technology, and in particular to a method and apparatus for determining family defects in power equipment. Background Technology

[0002] Family defects refer to the occurrence of the same type of defect in different models, specifications, and series of similar power equipment manufactured by the same company during operation. This defect may be caused by factors such as identical manufacturing processes, materials, design concepts, and approaches. In power grid systems, the reliability of power equipment directly affects the safe operation of the power system. Some equipment suffers from family defects due to factors such as design, materials, and manufacturing, resulting in a significantly higher failure rate after commissioning.

[0003] In current technology, the diagnosis of familial defects mainly relies on the subjective judgment of experts. However, this method of judgment lacks objective and quantitative unified standards, making it difficult to guarantee the accuracy and consistency of the diagnostic results.

[0004] Therefore, there is an urgent need for a method that can break away from the over-reliance on expert subjective judgment and establish objective and unified diagnostic criteria to more accurately determine whether electrical equipment has familial defects. Summary of the Invention

[0005] This application provides a method and apparatus for determining family defects in power equipment, with the aim of establishing an objective and unified judgment standard to more accurately determine whether power equipment has family defects.

[0006] To address the aforementioned technical problems, this application provides the following technical solutions:

[0007] In a first aspect, this application provides a method for determining family-related defects in electrical equipment, the method comprising:

[0008] Historical family-based defect data and corresponding target features are obtained for power equipment based on target categories;

[0009] Based on the target features, a preset knowledge graph model is traversed to determine the frequency of occurrence of the target features. The preset knowledge graph model is a correlation model constructed based on the operating data of the target power equipment to characterize the defect events of the power equipment.

[0010] Based on the historical family defect data, a feature analysis algorithm is used to determine the weight coefficients of the target features;

[0011] Based on the occurrence frequency and the weighting coefficient, the suspected familial defect rate is calculated to determine whether the target power equipment has a familial defect.

[0012] Secondly, this application provides an apparatus for determining family-related defects in electrical equipment, the apparatus comprising:

[0013] The acquisition unit is used to acquire historical family-based defect data and corresponding target features of power equipment based on the target category;

[0014] The determination unit is used to traverse a preset knowledge graph model based on the target features of the acquisition unit to determine the occurrence frequency of the target features. The preset knowledge graph model is a correlation model constructed based on the operating data of the target power equipment to characterize the defect events of the power equipment.

[0015] A weighting unit is determined to use a feature analysis algorithm based on the historical family defect data to determine the weighting coefficients of the target features.

[0016] The calculation unit is used to calculate the suspected family defect rate based on the occurrence frequency of the determination unit and the weight coefficient of the determination weight unit, so as to determine whether the target power equipment has a family defect.

[0017] Thirdly, this application provides a storage medium for storing a computer program, wherein the computer program, when running, controls the device where the storage medium is located to execute the above-described method for determining family defects in power equipment.

[0018] Fourthly, this application provides an electronic device including a processor and a memory, wherein the processor is configured to invoke program instructions in the memory to execute the above-described method for determining family defects in power equipment.

[0019] Compared to existing technologies, this technical solution firstly establishes objective diagnostic standards through quantitative feature analysis. Specifically, it extracts target features based on historical familial defect data and uses a knowledge graph model to statistically analyze the frequency of occurrence of these features, transforming qualitative judgments previously reliant on expert experience into quantifiable frequency indicators and avoiding subjective bias. Secondly, it assigns differentiated weight coefficients to different target features using feature analysis algorithms to reflect the contribution of each feature to familial defects, further improving diagnostic accuracy. Finally, it calculates the suspected familial defect rate based on occurrence frequency and weight coefficients, forming a complete objective and quantitative diagnostic system. This system can more accurately determine whether power equipment exhibits familial defects, improving the accuracy and consistency of diagnostic results and providing strong support for the safe and stable operation of the power grid system. Attached Figure Description

[0020] The above and other objects, features, and advantages of exemplary embodiments of this application will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of this application are illustrated by way of example and not limitation, with the same or corresponding reference numerals denoteing the same or corresponding parts, wherein:

[0021] Figure 1 A flowchart illustrating a method for determining family defects in electrical equipment according to an embodiment of this application is shown.

[0022] Figure 2 A flowchart of another method for determining family defects in electrical equipment according to an embodiment of this application is shown;

[0023] Figure 3 This application illustrates a flowchart of how to update a dictionary of faulty components or faulty locations, according to one embodiment of the present application.

[0024] Figure 4 A schematic diagram of a device structure for determining family defects in electrical equipment is shown in one embodiment of this application;

[0025] Figure 5 This illustration shows a schematic diagram of another device structure for determining family defects in electrical equipment, provided in another embodiment of this application. Detailed Implementation

[0026] Exemplary embodiments of this application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of this application and to fully convey the scope of this application to those skilled in the art.

[0027] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application shall have the ordinary meaning as understood by one of ordinary skill in the art to which this application pertains.

[0028] Currently, family defects refer to the occurrence of the same type of defect in different models, specifications, and series of electrical equipment manufactured by the same company, during operation. These defects may be caused by factors such as identical manufacturing processes, materials, and design concepts. In power grid systems, the reliability of electrical equipment directly affects the safe operation of the power system. Some equipment exhibits family defects due to design, materials, and manufacturing factors, leading to a significantly higher failure rate after commissioning. Other defects are frequently caused by climate, geographical environment, and human operation. Existing technologies for diagnosing family defects mainly rely on subjective expert judgment; however, this method lacks objective and quantitative unified standards, making it difficult to guarantee the accuracy and consistency of diagnostic results.

[0029] To address this issue, the inventors of this application propose a method for determining family defects in power equipment. This method, based on objective and unified diagnostic standards, can more accurately determine whether power equipment exhibits family defects. Specifically, it involves acquiring historical family defect data and target features of power equipment within that data; determining the frequency of occurrence of these target features using a pre-set knowledge graph model, where the model simulates the relationship between physical equipment information and operational event information within a power grid system; determining the weighting coefficients of the target features using a feature analysis algorithm based on the historical family defect data; and calculating the suspected family defect rate based on the frequency of occurrence and the weighting coefficients to confirm the presence of family defects in the power equipment. Therefore, this technical solution eliminates excessive reliance on subjective expert judgment, establishing objective and unified diagnostic standards for more accurate determination of whether power equipment exhibits family defects. The specific steps of a method for determining family defects in power equipment according to an embodiment of this application are as follows: Figure 1 As shown, it includes:

[0030] Step 101: Obtain historical family-based defect data and corresponding target features for power equipment based on the target category.

[0031] In this step, historical family defect data refers to relevant data on defects occurring in the same category of power equipment. This includes at least the equipment model, manufacturer, site, years of operation, and fault and defect events. The fault time and defect event specifically include information such as defect level, defect type, defect cause, defect location, and maintenance measures. Target features are key attributes extracted from historical family defect data that reflect the family-related defects of the equipment, such as manufacturer, equipment model, defective component, and defect cause. The target category refers to a specific type of power equipment; for example, all water pumps could be a target category of power equipment. If there are multiple existing power equipment categories, they could include transformers, switches, water pumps, etc.

[0032] In this step, historical family defect data can be obtained from multiple data sources within the power system, such as production management systems, equipment operation and maintenance records, and fault reports. This can be achieved through querying and filtering based on conditions such as time range and equipment type. Specifically, this could involve obtaining the operation and maintenance records of all transformers from 2021 to 2024. Data can be retrieved based on data time and equipment type within the data source. In this embodiment, the target features can be manually defined, or the historical family defect data can be analyzed and processed. Specifically, the Boruta feature algorithm can be used with data mining techniques to identify frequently occurring target features in family defects, such as features related to equipment attributes, fault modes, and operating environment.

[0033] Step 102: Based on the target features, traverse the preset knowledge graph model to determine the frequency of occurrence of the target features.

[0034] In this step, the preset knowledge graph model is a relational model built based on the operational data of the target power equipment to characterize defect events occurring in the power equipment. The preset knowledge graph model contains relationships between entities. Entities include physical equipment information and operational event information. Operational events include defect events and / or fault events. That is, the preset knowledge graph model stores the relationships between physical equipment information and operational event information, as well as the relationships between them. Operational events can be further subdivided into fault events and defect events, specifically including information such as defect level, defect type, defect cause, defect location, and maintenance measures. Physical equipment information includes equipment type, equipment model, manufacturer, site, and years of operation. In this embodiment, the target features are manufacturer, equipment model, defective component, and defect cause. The preset knowledge graph model is used to retrieve target feature information, i.e., to retrieve information from multiple dimensions such as equipment type, equipment model, and manufacturer, to determine the frequency of occurrence of the target features (manufacturer, equipment model, defective component, and defect cause information). When retrieving target feature information, after determining the category of target power equipment, for any target power equipment, fault events and defect events are sequentially retrieved in the preset knowledge graph model according to the order in which the target features appear. The retrieved information is marked, and the retrieved target feature data is recorded. Recording the retrieved target feature data can include the number of times the target feature appears, the specific information corresponding to the target feature, or a combination of both; no limitation is made here. A target power equipment can be a group of power equipment, indicating at least two units, or it can be a single unit; no further limitation is made here. Target power equipment refers to power equipment within the target category in the current power system.

[0035] In this step, the rules for matching target features with the preset knowledge graph can be the Rete rule matching algorithm. During the rule matching process, the rule conditions can be a rule engine, specifically a suspected family defect rule base. This rule base contains rules such as the generating manufacturer, equipment model, and problematic component. The rules in the suspected family defect rule base within the rule engine are updated using feature analysis. Specifically, this involves fault / defect feature analysis, followed by feature selection using the Boruta algorithm, and then rule evaluation to complete the rule update.

[0036] After determining the target features in step 101, the preset knowledge graph model is traversed based on the target features. At this point, the preset knowledge graph model traverses the target feature data one by one, recording the frequency of the target features during the traversal and marking the nodes where the target features appear to distinguish the traversed node information. In this step, during the traversal based on the target features, the number of times the target features appear is recorded. There are two ways to record the number of times: first, using a discrete recording method, that is, recording each time the target feature appears, and summing all the records after the traversal is completed; second, using an incremental recording method, that is, during the traversal, the recorded number is incremented every time the target feature is encountered.

[0037] Step 103: Based on the historical family defect data, use a feature analysis algorithm to determine the weight coefficients of the target features.

[0038] In this step, the Boruta feature analysis algorithm is used. The method for determining the weight coefficients of the target features is as follows: Shadow features are generated using the original features. A random forest model is trained using the original and shadow features. The random forest model calculates the importance score (weight coefficient) of each feature (original and shadow features). The importance score of each original feature is compared with the maximum importance score of all shadow features. If the importance score of an original feature is consistently higher than the maximum importance score of a shadow feature, after multiple iterations, the original feature is considered "important" and retained; otherwise, it is removed. Based on the target feature importance ranking obtained from the Boruta feature analysis algorithm and the results of manual adjustment, the retained features are weighted. A normalization method is used to ensure that the weights of each target feature are between 0 and 1, and the sum of all weights is 1.

[0039] Here is an example: We have collected 100 published cases of family defects in power transformers. Each case includes information such as the manufacturer, equipment model, defective component, cause of defect, type of defect, and whether it is a family defect.

[0040] The order of data within the "Manufacturer" feature is randomly shuffled to obtain corresponding shadow features. For example, under the "Manufacturer" feature, there are Manufacturer 1, Manufacturer 2, and Manufacturer 3. The order of Manufacturer 1, Manufacturer 2, and Manufacturer 3 is randomly shuffled to obtain the corresponding shadow features. A random forest algorithm is used to train the dataset containing the original features and shadow features to calculate the importance score of each target feature. For example, the importance score of Manufacturer is 0.3, Equipment Model is 0.25, Defective Component is 0.2, Defect Cause is 0.2, and Defect Type is 0.05. After multiple iterations and comparisons, it is found that the importance scores of Manufacturer, Equipment Model, Defect Cause, and Defective Component are consistently higher than the maximum importance score of the shadow features, while the importance score of Fault Type is lower than the maximum importance score of the shadow features. Therefore, the Fault Type feature is removed. The target features are now: Manufacturer, Equipment Model, Defective Component, and Defect Cause. The weights of each target feature are determined based on historical family defects. If, based on empirical values, the defective component is determined to be more important in judging family defects, its weight is increased. That is, the final weights are determined as follows: the importance score of the manufacturer is 0.3, the importance score of the equipment model is 0.25, the importance score of the defective component is 0.2, and the importance score of the cause of the defect is 0.25.

[0041] Step 104: Calculate the suspected familial defect rate based on the occurrence frequency and the weighting coefficient to determine whether the power equipment has a familial defect.

[0042] In this step, based on the frequency of occurrence in step 102 and the weighting coefficients in step 103, the probability of suspected family defects is calculated. The specific calculation method is as follows:

[0043] P = A1 × B1 + … + A n ×B n

[0044]

[0045] Where P is the probability of suspected family defects, A is the weight coefficient of the target feature, n is the number of target features, B is the frequency of any target feature, and b represents the frequency of any target feature in the preset knowledge graph model.

[0046] In this embodiment, after determining the suspected family defect rate of power equipment according to the above method, the suspected family defect rates are sorted in reverse order, and it is determined whether the first-ranked suspected family defect rate is greater than a preset probability. If so, a family defect is confirmed. At this time, the unique identifier and location information of the power equipment with family defects are determined in the preset knowledge graph model. The system then sends the unique identifier and location information of the power equipment to the display screen, which is used to remind maintenance personnel to pay attention to the family defect fault of the power equipment, thereby realizing the discovery and early warning of family defects. For example, the target category in the current scenario is: switch components. The current power system has switch components 1, 2, and 3. The suspected family defect rates of the above switch components are calculated respectively, and the suspected family defect rates are sorted in reverse order. It is determined whether the first-ranked suspected family defect rate is greater than a preset probability. If so, the target equipment is confirmed to have a family defect. The preset probability is a probability value preset based on the actual situation, and the specific value of the preset probability is not limited here. If the first-ranked suspected family defect rate is less than the preset probability, it means that the target power equipment does not have a family defect. If the highest-ranked suspected family defect rate is greater than a preset probability, then the second-ranked suspected family defect rate is checked against the preset probability, using the same method, until the determination of whether the target device has a family defect event is completed. Alternatively, the suspected family defect rates can be sorted from smallest to largest during the judgment process; the specific judgment mode is the same as for reverse sorting, and will not be elaborated here. In this embodiment, if the power equipment switching component is only switch component 1, then no sorting is required, and the judgment can be performed directly.

[0047] Furthermore, based on the above Figure 1 The embodiments of the present invention shown provide a more detailed explanation of how to construct a preset knowledge graph model, with specific steps as follows: Figure 2 As shown, it includes:

[0048] Step 201: Obtain the operating data of the target power equipment.

[0049] In this step, the operational data of the target power equipment includes physical equipment information and operational event information, that is, all operational data of the target power equipment whose familial defects need to be determined. All operational data represents operational event information and physical equipment information that has occurred with the power equipment. Physical equipment information includes equipment type, model, manufacturer, site, and years of operation. Operational events can be further subdivided into fault events and defect events, specifically including defect level, defect type, defect cause, defect location, and maintenance measures. The power equipment can be of various types, such as generators, transformers, transmission lines, switchgear, and related control, protection, and monitoring devices. The power equipment data can be collected from multiple different data sources, such as the power company's equipment management system (recording basic equipment information and maintenance records), SCADA (Supervisory Control and Data Acquisition) system (providing real-time equipment operational data), fault reporting system (recording fault-related data), and equipment ledgers.

[0050] Step 202: Based on the physical device identifiers and operation event identifiers in the operation data, as well as the relationship between physical devices and operation events, construct the framework of a preset knowledge graph model.

[0051] In this step, physical device identifiers and operational event identifiers are obtained from the operational data. A physical device identifier is a unique identifier for a device within the power system, such as generator 001 or switchgear 100. An operational event identifier refers to a fault event or defect event. For the current power system, if there are 10 power devices, 10 preset knowledge graph model frameworks are constructed; if there are 100 power devices, 100 preset knowledge graph frameworks are constructed.

[0052] In this step, the framework for constructing the pre-defined knowledge graph model can be established by first identifying the physical device identification information in the operational data to create entity nodes. Then, based on the physical device identification information, operational events related to the physical devices are searched to identify the operational event identifiers and designate them as attribute nodes. Finally, the relationship between the physical device and the operational event is added between the two nodes to establish the framework of the pre-defined knowledge graph model. For example, if an event ID occurs when a device ID occurs, the device ID can be used as the entity node, the event ID as the attribute node, and "occurred" as the connection between the two to establish the framework.

[0053] Step 203: Based on the equipment fault and defect analysis data in the operation data, render the framework of the preset knowledge graph model to obtain the preset knowledge graph model.

[0054] In this step, the equipment fault and defect analysis data consists of detailed records and analyses of faults or defects occurring in power equipment. This includes descriptions of the fault phenomena, the specific location of the fault, diagnostic analysis of the fault causes, measures taken to resolve the fault, and related inspection reports and data. The equipment fault and defect analysis data is textual information such as equipment fault and defect analysis reports. In this step, the rendering of the framework is based on data extracted from the equipment fault and defect analysis data. The extracted data is added to the framework to determine the preset knowledge graph model.

[0055] In this step, entity information of the preset knowledge graph model can be obtained from the equipment fault defect analysis data using a progressive dictionary text extraction method, a rule-based extraction method, or a regular expression extraction method. In this embodiment, the progressive dictionary text extraction method is used, as detailed below:

[0056] Scenario 1 (Equipment Dimension): Based on the standard terminology for power transmission and transformation equipment defects, generate dictionaries corresponding to equipment type, equipment model, manufacturer, fault location, and fault component; after preprocessing the equipment fault defect analysis data, extract the equipment type, equipment model, manufacturer, fault location, and fault component, and import them into the corresponding dictionaries to determine the first dictionary and the second dictionary. The first dictionary includes an equipment type dictionary, an equipment model dictionary, and a manufacturer dictionary; based on the long text of cause analysis in the equipment fault defect analysis data, sequentially traverse the information in the first dictionary to determine the first entity data of the preset knowledge graph model; based on the first entity data, render the framework of the preset knowledge graph model to obtain the preset knowledge graph model.

[0057] Specifically: The Standard for Terminology of Power Transmission and Transformation Equipment Defects is a standardized language system for describing and defining various defects that occur during the operation of power transmission and transformation equipment. It specifies unified terminology, definitions, and expressions to ensure accurate and consistent descriptions of equipment defects. The Standard for Terminology of Power Transmission and Transformation Equipment Defects is obtained from equipment ledger information in the Production Management System (PMS) and structured data from the State Grid Corporation of China's enterprise standard: "Q / GDW1904.1-2013 Standard for Terminology of Power Transmission and Transformation Equipment Defects Part 1: Primary Substation Part". The Equipment Type Dictionary is a collection of terms specifically storing different types of power equipment. It is used to classify and identify power equipment, facilitating accurate identification and management of equipment type information in knowledge graphs or related data processing. Through the Equipment Type Dictionary, the category to which equipment belongs can be quickly determined, such as transformers, circuit breakers, transmission lines, etc. The Equipment Model Dictionary mainly stores terminology information for specific models of various power equipment. Equipment models include detailed characteristics such as technical parameters and specifications; different models represent equipment with different performance and features. The equipment model dictionary helps to accurately distinguish and describe specific equipment, serving as a crucial basis for detailed equipment identification and management within the knowledge graph. The manufacturer dictionary is a collection of terms recording the names of various manufacturers producing power equipment. Preprocessing of equipment fault and defect analysis data involves removing all special characters and punctuation marks from the data. The long text of the cause analysis is text content that provides a detailed analysis and description of the causes of a specific equipment fault or defect, typically including a detailed explanation of the fault phenomenon, possible cause speculations, and related technical analysis. The first entity data is entity data related to equipment type, equipment model, and manufacturer, extracted from the first dictionary by processing the long text of the cause analysis in the equipment fault and defect analysis data. This data will be used to populate the knowledge graph framework. During traversal, string matching algorithms (such as exact matching and fuzzy matching) can be used to search for identical words in the long text of the cause analysis compared to those in the first dictionary. Rendering the framework of the preset knowledge graph model uses knowledge graph building tools (such as Neo4j, Graphviz, etc.) or programming languages ​​(such as Python combined with drivers for relevant graph databases) to load the framework of the preset knowledge graph model.

[0058] After identifying the first entity data, a specific label is assigned to it, which can be a specific color, symbol, or code. The purpose of the label is to classify, differentiate, or convey specific information about the entity data, facilitating subsequent visualization, querying, and analysis of the knowledge graph. If the word data in the long text of the cause analysis is not present in the first dictionary, the long text of the cause analysis is manually obtained and the first data is extracted. This first data represents the equipment type, model, and manufacturer of the power equipment and is added to the information in the first dictionary to ensure the accuracy of the information in the first dictionary.

[0059] Scenario 2 (Production Dimension): Segment the short phrases indicating fault causes in the equipment fault defect analysis data to obtain a segmentation list; match the words in the segmentation list with the fault location dictionary and the fault component dictionary to determine the second entity data of the preset knowledge graph model; based on the second entity data, render the framework of the preset knowledge graph model to obtain the preset knowledge graph model.

[0060] Specifically: The first part of the sentence describes the cause of a fault in the equipment fault defect analysis data. For example, a sentence like "The equipment failed due to a short circuit in the transformer winding." Word segmentation involves dividing a sentence or text into individual words according to certain rules for subsequent processing and analysis. This can be done using Jieba segmentation; for instance, segmenting "The equipment failed due to a short circuit in the transformer winding" yields "due to," "transformer," "winding," "short circuit," "caused," "equipment," and "fault." The segmentation can be done by importing the data into a word segmenter. After segmenting the fault cause sentence, each word and its part of speech are read to obtain a word list. This list can be a set of words ordered by their part of speech or directly ordered by the order in which the words appear in the fault cause sentence. The second entity data consists of vocabulary related to the fault location and fault component, obtained by matching the words in the word list with a fault location dictionary and a fault component dictionary. The fault location dictionary is a collection of vocabulary specifically used to store the specific locations where problems might occur when power equipment malfunctions. Example: For transformer equipment, a fault location dictionary might include "winding parts," "core parts," "bushing parts," and "tap switch parts," etc.; for transmission lines, it might include "tower parts," "insulator string parts," and "conductor connection parts," etc.; for circuit breakers, it might include "contact parts," "arc-extinguishing chamber parts," and "operating mechanism parts," etc. A fault component dictionary, on the other hand, is a collection of terms related to specific faulty parts in power equipment. Unlike a fault location dictionary, which focuses on location, a fault component dictionary focuses more on specific parts or components within the equipment. When equipment malfunctions, the fault component dictionary can clearly identify which specific component caused the fault, which is crucial for fault diagnosis, repair, and prevention. It provides detailed information about the faulty object for equipment maintenance and management. Example: In transformers, a fault component dictionary might include "winding coils," "core silicon steel sheets," "bushing porcelain insulators," and "tap switch contacts," etc.; in transmission lines, it includes "insulators," "line clamps," and "vibration dampers," etc.; in circuit breakers, it includes "moving and stationary contacts," "arc-extinguishing chamber nozzles," and "operating mechanism springs," etc.

[0061] For example, if an event ID occurs for a device ID, then the device ID can be used as an entity node, the event ID as an attribute node, and "occurred" as the connection between them, establishing a framework T. The second entity data extracted from the device fault analysis data is defect level A, and the first entity data extracted is the device manufacturer C. Then, A and C are rendered on framework T. It can be determined that manufacturer C is an attribute node of the device ID. The relationship between the device ID and the device manufacturer (an inherent attribute of the device) is an inclusion relationship. Therefore, device manufacturer C can be rendered into framework T. After rendering device manufacturer C, the relationship between defect level A and event ID is found; this is a connection relationship. Defect level is an attribute node of event ID. The connection relationship and attribute node A are rendered into framework T containing device manufacturer C, completing the rendering of the preset knowledge graph model. It is worth noting that this example only uses defect level and device manufacturer for illustration. In actual operation, attribute nodes can be information such as defective component, defective location, defective event, defect type, defect cause, defect level, maintenance measures, equipment model, voltage level, years of operation, equipment type, and manufacturer, etc., without further limitations.

[0062] In the above step of "matching words in the word segmentation list with the fault location dictionary and the fault component dictionary", the fault component dictionary or fault location dictionary is also updated. The specific implementation method is as follows: Figure 3 Specifically, the process is as follows: A short text describing the cause of the fault is obtained. This text is segmented using Jieba, and words and their parts of speech are read sequentially. It is determined whether a word is a noun; if so, it is added to a candidate list. If not, it is imported into the fault location dictionary and the fault component dictionary (to determine which dictionary the word belongs to). After determining that a word is a noun, a candidate list is obtained, consisting of two words with noun parts of speech in the fault cause sentence. This candidate list is then matched against the fault location dictionary and the fault component dictionary. The matching order is based on the condition that only one word successfully matches either the fault location dictionary or the fault component dictionary. If only one word successfully matches either dictionary, the fault component dictionary or the fault location dictionary is updated, and the successfully matched word is tagged to confirm a successful match. If the words in the segmented list do not match either the fault location dictionary or the fault component dictionary, the fault cause sentence is manually retrieved, and second data is extracted. This second data represents words related to the fault location and fault component of the power equipment and is added to the second dictionary to ensure its accuracy.

[0063] In this embodiment, after determining the first entity data and the second entity data, a preset knowledge graph model is generated using the Neo4j graph database, specifically as follows: The Neo4j graph database generates entity nodes (physical device identifier entity nodes and runtime event identifier entity nodes) and connection relationships (relationships between physical devices and runtime events) based on the framework in step 202. The framework creation process involves first rendering the physical device identifier entity nodes, then rendering the connection relationships, and finally rendering the runtime event identifier entity nodes. Based on this framework, attribute nodes (first entity data and second entity data) and connection relationships between attribute nodes and entity nodes are created one by one. Specifically, attribute nodes (first entity data) that are connected to the physical device identifier entity node are added to it. If the first entity data contains only one attribute node, the first entity data is directly added to the physical device identifier entity node. If the first entity data also has child nodes, child nodes are added according to the connection relationships between their child nodes and attribute nodes, and so on, until all nodes in the first entity data have been added. If there are multiple child nodes, for nodes of the same level, child nodes without leaf nodes are added first. Furthermore, based on the aforementioned framework, attribute nodes (second entity data) are added to the runtime event identifier entity nodes. The second entity data is then added to the runtime event identifier entity nodes. If the attribute node (second entity data) has child nodes, these child nodes are added according to their connection relationships with the attribute node, and so on, until all nodes in the second entity data are added. If there are multiple child nodes, for nodes of the same level, child nodes without leaf nodes are added first to complete the rendering of the preset knowledge graph model. Alternatively, when rendering the framework, entity nodes and attribute nodes can be rendered first, and then the preset knowledge graph model can be rendered based on the connection relationships. Specifically, when rendering entity and attribute nodes, entity nodes should be rendered first, followed by attribute nodes.

[0064] Furthermore, attributes such as images and videos used for recording during inspections and live-line testing are added to event-type entities in the pre-defined knowledge graph model. Address information for images and videos is stored using a relational database, with the attribute ID from the graph database serving as the unique primary key for the entire text. This primary key information is then stored as an attribute of the event-type entity, achieving the associated storage of the knowledge graph and related audio files.

[0065] Furthermore, to ensure the standardized integration of the first and second data, knowledge fusion is performed on the specific content of the first and / or second data, that is, the meaning of the same description is normalized to achieve attribute co-reference resolution. First, the contents of the first and second dictionaries are sorted according to dictionary order, and the data in the dictionaries are initially classified, that is, sorted according to the 26 English letters, and then the entity data is standardized and integrated. The specific steps are as follows: obtain the attribute values ​​of the first and / or second data, which are composed of numerical values ​​and text; calculate the numerical similarity and text similarity based on the numerical values ​​and text; determine the similarity of entity attributes based on the numerical similarity and text similarity; and perform knowledge fusion on the first and / or second data based on the similarity of entity attributes. Entity attribute values ​​are usually composed of numerical values ​​and text, so numerical similarity and text similarity can be calculated separately, and the final entity attribute similarity S(A, B) can be calculated by weighting.

[0066] S(A, B) = αS i (A,B)+βS j (A,B)

[0067] Where A and B are two entity attribute values; S i (A, B) represents numerical similarity, S j (A, B) represents textual similarity; α is the numerical weight, β is the textual weight, and α + β = 1.

[0068] The formulas for calculating the weights of numerical and textual data are as follows:

[0069]

[0070] Among them, L m L is the length of the numerical type. t L represents the length of text attributes, and L represents the total length of entity attributes. The length of numeric attributes refers to the number of numeric attributes within the entity attributes. For example, for an electrical equipment entity, its attributes might include the equipment's rated voltage (numerical), rated current (numerical), and operating temperature (numerical). If this entity has 3 numeric attributes, then L... m =3, that is to say, L m This counts the number of attributes represented in numerical form that an entity contains. The length of a text attribute refers to the number of text attributes within the entity's attributes. For example, the "electrical equipment" entity mentioned above might also have attributes such as "equipment model" (text), "manufacturer" (text), and "installation location" (text). If there are three text attributes, then L... t =3, that is, L t It counts the number of attributes represented in text form for the entity.

[0071] For numeric attributes, directly check if the two values ​​are the same; if they are the same, then S. i (A, B) = 1, if they are inconsistent, then S i (A, B) = 0.

[0072] For text-based attributes, the word2vec word vector model is used to convert the attribute values ​​to be calculated into word vectors. Similarity is then calculated using cosine similarity, with the specific formula as follows:

[0073]

[0074] Among them, A n Let B be an n-dimensional vector of attribute A. n Let n be the n-dimensional vector value of attribute B, where n represents the dimension of the word.

[0075] The similarity score for entity attributes ranges from 0 to 1. The higher the similarity score, the greater the similarity between the two entity attributes. When the similarity score exceeds a preset threshold, it can be preliminarily determined that the two entity attributes can be merged to achieve unified classification of entity attributes. After manual verification, it is finally confirmed whether merging is necessary to achieve unified classification of entity attributes and thus realize knowledge fusion.

[0076] Here are some examples to illustrate the above:

[0077] Given: Entity A has the following equipment model (text attribute): "S11-1000 / 10", rated capacity (numerical attribute): 1000 (kVA), and manufacturer (text attribute): "NARI Electric Co., Ltd."; Entity B has the following equipment model (text attribute): "S11-1000 / 10", rated capacity (numerical attribute): 1000 (kVA), and manufacturer (text attribute): "NARI Electric Co., Ltd.", with a preset threshold of 0.8.

[0078] First, determine the lengths and weights. Numerical attributes include the rated capacity L. m =1, text-type attributes include device model and manufacturer L t =2, the total length of entity attributes is 1+2=3, then the numerical weights α=1 / 3, β=2 / 3. Next, numerical similarity is determined. Since the rated capacity of both entity A and entity B is 1000kvA, S i (A, B) = 1. The text similarity is recalculated as follows:

[0079] 1. Equipment Model

[0080] Suppose we have already trained a word vector model using word2vec on a large amount of text data related to power equipment. For the equipment model "S11-1000 / 10", we convert it into a word vector. Let's assume the converted word vector for entity A's equipment model is: A model = [a1, a2, ..., an], the word vector for the device model of B is: B model = [b1, b2, ..., bn], where the equipment models are exactly the same, S model =1.

[0081] 2. Manufacturer

[0082] Convert the manufacturer of entity A, "NARI Electric Co., Ltd.", and the manufacturer of entity B, "NARI Electric Co., Ltd.", into word vectors. The word vector for the equipment model of entity A is: A model =[c1, c2, ..., cn], the word vector for the device model of B is: B model = [d1, d2, ..., dn]. If S is calculated according to the above formula... model =0.9, at this time S j (A, B) = (1 + 0.9) / 2 = 0.95. Finally, the entity attribute similarity S(A, B) = 0.9665. Since 0.9665 is greater than 0.8, it is initially determined that these two entity attributes can be merged. After manual verification, it was found that the two entities do indeed describe the same type of transformer, only with slight differences in the manufacturer's name. Ultimately, it was confirmed that these two entities would be merged, unifying the descriptions of equipment model, rated capacity, and manufacturer, achieving unified classification of entity attributes, and thus completing knowledge fusion.

[0083] Furthermore, in this embodiment, the normalization processing of the first and second data can also involve adding them to the first and second dictionaries respectively after acquisition, extracting them from the dictionaries, and sorting the contents of the first and second dictionaries. The sorting method can be based on the alphabetical order of the first character of each word in the dictionary content, grouping words with the same first letter as a single word and performing a simple sort. Then, knowledge fusion is completed according to the calculation method described above. If, after acquiring the first and second entity data, normalization processing is performed on the acquired first and second entity data, the above method can also be used. That is, data content in the first and / or second entity data that refers to different names of the same thing is normalized in the manner described above.

[0084] Furthermore, as a response to the above Figure 1-3The implementation of the method embodiment shown in this invention provides a device for determining family defects in power equipment. This device is used to establish objective and unified judgment criteria to more accurately determine whether power equipment has family defects. The embodiment of this device corresponds to the foregoing method embodiment. For ease of reading, this embodiment will not repeat the details of the foregoing method embodiment, but it should be clear that the device in this embodiment can implement all the contents of the foregoing method embodiment. Specifically, as shown... Figure 4 As shown, the device includes: an apparatus for determining a family-related defect in electrical equipment, the apparatus comprising:

[0085] The acquisition unit 41 is used to acquire historical family-based defect data and corresponding target features of power equipment based on the target category;

[0086] The determining unit 42 is used to traverse a preset knowledge graph model based on the target features of the acquiring unit 41 to determine the occurrence frequency of the target features. The preset knowledge graph model is a correlation model constructed based on the operating data of the target power equipment to characterize the defect events of the power equipment.

[0087] The weighting unit 43 is used to determine the weight coefficient of the target feature based on the historical family defect data of the acquisition unit 41 using a feature analysis algorithm.

[0088] The calculation unit 44 is used to calculate the suspected familial defect rate based on the occurrence frequency of the determination unit 42 and the weighting coefficient of the determination weighting unit 43, so as to determine whether the target power equipment has a familial defect. Further, as... Figure 5 As shown, the device further includes a model building unit 45, which includes:

[0089] Data acquisition module 451 is used to acquire the operating data of the target power equipment;

[0090] The construction module 452 is used to construct a framework for a preset knowledge graph model based on the physical device identifier and operation event identifier in the operation data of the data acquisition module 451 and the relationship between physical devices and operation events. The operation events include at least defect event information and fault event information.

[0091] The rendering module 453 is used to render the framework of the preset knowledge graph model of the construction module 452 based on the equipment fault and defect analysis data in the running data of the data acquisition module 451, so as to obtain the preset knowledge graph model.

[0092] Furthermore, such as Figure 5 As shown, the rendering module 453 includes:

[0093] The generation submodule 4531 is used to generate dictionaries corresponding to equipment type, equipment model, manufacturer, fault location, and fault component based on the standard terminology for defects in power transmission and transformation equipment.

[0094] The dictionary submodule 4532 is used to preprocess the equipment fault defect analysis data of the generation submodule 4531, extract the equipment type, equipment model, manufacturer, fault location and fault component, and import them into the corresponding dictionary to determine the first dictionary and the second dictionary. The first dictionary includes an equipment type dictionary, an equipment model dictionary and a manufacturer dictionary.

[0095] The entity data determination submodule 4533 is used to determine the first entity data of the preset knowledge graph model by sequentially traversing the information in the first dictionary of the dictionary determination submodule 4532 based on the cause analysis long text in the equipment fault defect analysis data.

[0096] The rendering framework submodule 4534 is used to render the framework of the preset knowledge graph model based on the first entity data of the determined entity data submodule 4533, so as to obtain the preset knowledge graph model.

[0097] Furthermore, such as Figure 5 As shown, the second dictionary includes a fault location dictionary and a fault component dictionary, and the rendering module 453 includes:

[0098] The submodule 4535 is used to segment the fault cause phrases in the equipment fault defect analysis data to obtain a segmentation list.

[0099] The matching submodule 4536 is used to match words in the word segmentation list of the acquisition list submodule 4535 with the fault location dictionary and the fault component dictionary to determine the second entity data of the preset knowledge graph model.

[0100] The rendering graph submodule 4537 is used to render the framework of the preset knowledge graph model based on the second entity data of the matching submodule 4536, so as to obtain the preset knowledge graph model.

[0101] Furthermore, such as Figure 5 As shown, the matching submodule 4536 includes: obtaining a candidate list, wherein the candidate list consists of words with noun parts of speech in the short phrase of fault cause, and the number of words is two; matching the candidate list with the fault location dictionary and the fault component dictionary respectively; if only one word is successfully matched with the fault location dictionary or the fault component dictionary, then updating the fault component dictionary or the fault location dictionary.

[0102] Furthermore, such as Figure 5 As shown, the adding module 454 includes:

[0103] The first addition submodule 4541 is used to manually obtain the long text of cause analysis and extract the first data if the word data in the long text of cause analysis does not exist in the information of the first dictionary. The first data represents the words of equipment type, equipment model and manufacturer of power equipment, and is used to add them to the information of the first dictionary.

[0104] The second addition submodule 4542 is used to manually obtain short sentences of fault causes and extract second data if the words in the word segmentation list do not match the fault location dictionary and the fault component dictionary. The second data represents words of fault location and fault component of power equipment and is used to add information to the second dictionary.

[0105] Furthermore, such as Figure 5 As shown, before rendering the preset knowledge graph model to obtain the preset knowledge graph model, the device further includes a fusion module 455 comprising:

[0106] The attribute value acquisition submodule 4551 is used to acquire the attribute values ​​of the first data and / or the second data, wherein the attribute values ​​consist of numerical values ​​and text.

[0107] The similarity calculation submodule 4552 is used to calculate numerical similarity and text similarity based on the numerical value and text of the attribute value acquisition submodule 4551.

[0108] The similarity calculation submodule 4552 determines the similarity of entity attributes based on the numerical similarity and text similarity.

[0109] A fusion submodule 4553 is defined, which is used to perform knowledge fusion on the first data and / or the second data based on the similarity of the entity attributes of the similarity calculation submodule 4552.

[0110] Furthermore, embodiments of this application also provide a computing device, the computing device comprising: at least one processor, and a memory, wherein the memory stores instructions executable by the processor, the instructions being executed by the processor, thereby enabling the processor to perform the above-described operations. Figure 1-3 The method for determining family defects in power equipment as described in [the document].

[0111] Furthermore, embodiments of this application also provide a readable storage medium for storing a computer program, wherein the computer program, when running, controls the device where the storage medium is located to perform the above-described actions. Figure 1-3 The method for determining family defects in power equipment as described in [the document].

[0112] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0113] It is understood that the relevant features in the above methods and apparatus can be referenced interchangeably. Furthermore, the terms "first," "second," etc., in the above embodiments are used to distinguish between embodiments and do not represent the superiority or inferiority of any particular embodiment.

[0114] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0115] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, this invention is not directed to any particular programming language. It should be understood that the contents of the invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the invention.

[0116] In addition, the memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0117] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0118] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0119] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0120] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0121] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0122] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0123] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0124] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0125] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The above descriptions are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for determining family-related defects in electrical equipment, characterized in that, The method includes: Historical family-based defect data and corresponding target features are obtained for power equipment based on target categories; Based on the target features, a preset knowledge graph model is traversed to determine the frequency of occurrence of the target features. The preset knowledge graph model is a correlation model constructed based on the operating data of the target power equipment to characterize the defect events of the power equipment. Based on the historical family defect data, a feature analysis algorithm is used to determine the weight coefficients of the target features; Based on the occurrence frequency and the weighting coefficient, the suspected familial defect rate is calculated to determine whether the target power equipment has a familial defect.

2. The method according to claim 1, characterized in that, The method includes: Acquire the operating data of the target power equipment; Based on the physical device identifiers and operational event identifiers in the operational data, as well as the relationship between physical devices and operational events, a framework for a pre-defined knowledge graph model is constructed. The operational events include at least defect event information and fault event information. Based on the equipment fault and defect analysis data in the operational data, the framework of the preset knowledge graph model is rendered to obtain the preset knowledge graph model.

3. The method according to claim 2, characterized in that, The process of analyzing equipment faults and defects based on the data from the power equipment, rendering the framework of the preset knowledge graph model to obtain the preset knowledge graph model includes: Based on the standard terminology for defects in power transmission and transformation equipment, a dictionary is generated corresponding to equipment type, equipment model, manufacturer, fault location, and faulty component. After preprocessing the equipment fault defect analysis data, the equipment type, equipment model, manufacturer, fault location, and fault component are extracted and imported into the corresponding dictionaries to determine the first dictionary and the second dictionary. The first dictionary includes an equipment type dictionary, an equipment model dictionary, and a manufacturer dictionary. Based on the long text of cause analysis in the equipment fault defect analysis data, the information in the first dictionary is sequentially traversed to determine the first entity data of the preset knowledge graph model; Based on the first entity data, the framework of the preset knowledge graph model is rendered to obtain the preset knowledge graph model.

4. The method according to claim 3, characterized in that, The second dictionary includes a fault location dictionary and a fault component dictionary. The process of rendering the framework of the preset knowledge graph model based on equipment fault defect analysis data from the power equipment data to obtain the preset knowledge graph model includes: The fault cause phrases in the equipment fault defect analysis data are segmented into words to obtain a word segmentation list; The words in the word segmentation list are matched with the fault location dictionary and the fault component dictionary to determine the second entity data of the preset knowledge graph model. Based on the second entity data, the framework of the preset knowledge graph model is rendered to obtain the preset knowledge graph model.

5. The method according to claim 4, characterized in that, The step of matching words in the word segmentation list with the fault location dictionary and the fault component dictionary to determine the second entity data of the preset knowledge graph model also includes: Obtain a candidate list, which consists of two words with noun parts of speech in the short sentences describing the cause of the fault. The candidate list is matched with the fault location dictionary and the fault component dictionary, respectively; If only one word successfully matches the fault location dictionary or fault component dictionary, then update the fault component dictionary or fault location dictionary.

6. The method according to claim 4, characterized in that, Before rendering the preset knowledge graph model to obtain the preset knowledge graph model, the method further includes: If the information in the first dictionary does not contain the word data in the long text of cause analysis, then the long text of cause analysis is manually obtained and the first data is extracted. The first data represents the words of equipment type, equipment model and manufacturer of power equipment, which are used to add to the information in the first dictionary. If the words in the word segmentation list do not match the fault location dictionary and the fault component dictionary, then short sentences about the cause of the fault are manually obtained and second data is extracted. The second data represents words representing the fault location and fault component of the power equipment and is used to add information to the second dictionary.

7. The method according to claim 4, characterized in that, Before rendering the preset knowledge graph model to obtain the preset knowledge graph model, the method further includes: Obtain the attribute values ​​of the first data and / or the second data, wherein the attribute values ​​consist of numerical values ​​and text; Based on the numerical values ​​and text, calculate the numerical similarity and text similarity; Based on the numerical similarity and text similarity, the similarity of entity attributes is determined; Based on the similarity of the entity attributes, knowledge fusion is performed on the first data and / or the second data.

8. An apparatus for determining family-related defects in electrical equipment, characterized in that, The device includes: The acquisition unit is used to acquire historical family-based defect data and corresponding target features of power equipment based on the target category; The determination unit is used to traverse a preset knowledge graph model based on the target features of the acquisition unit to determine the occurrence frequency of the target features. The preset knowledge graph model is a correlation model constructed based on the operating data of the target power equipment to characterize the defect events of the power equipment. A weighting unit is determined, which is used to determine the weighting coefficient of the target feature based on the historical family defect data of the acquisition unit using a feature analysis algorithm; The calculation unit is used to calculate the suspected family defect rate based on the occurrence frequency of the determination unit and the weight coefficient of the determination weight unit, so as to determine whether the target power equipment has a family defect.

9. A storage medium, characterized in that, The storage medium is used to store a computer program, wherein, when the computer program is executed, it controls the device on which the storage medium is located to perform the method for determining family defects of power equipment as described in any one of claims 1-7.

10. An electronic device, characterized in that, The electronic device includes a processor and a memory, the processor being configured to invoke program instructions in the memory to execute the method for determining family defects in power equipment as described in any one of claims 1-7.

Images