A system and method for automatically repairing abnormal data in power grid equipment asset ledger

Abstract

A power grid equipment asset ledger abnormal data automatic repair system includes a power grid asset data management module, an abnormal data positioning and marking module, an abnormal data clustering analysis module, and an abnormal data automatic repair module. When the system is running, the power grid asset data management module obtains the original data information of the power grid equipment assets from the external system, and the abnormal data positioning and marking module then identifies and marks the abnormal fields in the original data, and determines the data abnormality type and associated fields corresponding to each abnormal field. Subsequently, the abnormal data clustering analysis module constructs a weighted clustering algorithm model based on the marked abnormal fields, and clusters the marked abnormal fields according to the data abnormality type to form multiple abnormal classification data sets. Finally, the abnormal data automatic repair module uses the corresponding repair scheme for each abnormal classification data set to repair the abnormal data. This design realizes the rapid, accurate, and reliable positioning detection and repair of abnormal data in the power grid equipment asset ledger.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a system and method for automatically repairing abnormal data in a power grid equipment asset ledger. Background Art

[0002] With the continuous improvement of the modern equipment management system of power grid enterprises, the number of physical assets put into operation has increased, and the amount of operation and maintenance data has gradually increased. It is inevitable that some abnormal situations will occur in the statistical data of the ledger, such as missing fields, abnormal coding, abnormal values, abnormal equipment status, etc. In addition, since the data comes from multiple management systems, there are often problems of inconsistent data between different data sources, such as the mismatch between the number of equipment and the total value of assets. These data anomalies directly affect the accuracy of power grid asset statistics, which further affects the scientific nature of decision-making. It can be seen that accurately locating and repairing abnormal data in the power grid ledger is of great practical significance for power grid planning, development and operation decisions.

[0003] In the early days, the method of on-site investigation was mostly used to detect abnormal data in power grid equipment ledgers, which was extremely labor-intensive and material-intensive and inefficient. With the development of detection technology, people have made some improvements in methods, but the technology is relatively backward and there is a problem of low detection accuracy. In recent years, relevant scholars have conducted relevant research on data anomaly detection of power grid equipment. The literature "Improved Data Anomaly Detection Method Based on Isolation Forest" (Xu Dong, Wang Yanjun, Meng Yulong, et al. [J]. Computer Science, 2018, 45(10): 155-159.) proposed a SA-iForest algorithm, which has high execution efficiency and accuracy in data anomaly detection, but is only suitable for anomaly detection of continuous data. For anomaly detection of non-continuous data, its accuracy and stability are poor. The paper “Time series anomaly detection method based on frequent pattern discovery” (Li Hailin, Wu Xianli. [J]. Computer Applications, 2018, 38(11): 3204-3210.) proposed a time series data anomaly detection method. This method analyzes the overall and local characteristics of the sequence. Although it can accurately find the abnormal fragments in the time series, it requires too many parameters to be set and the parameter adjustment process is complicated. The paper “New energy power data repair technology based on big data analysis” (Tian Kun. [J]. Urban Construction, 2018(11): 375-376.) uses big data thinking to deeply analyze the relevant laws of new energy power generation operation data mining and proposes a new energy power generation data anomaly repair technology, but the application value of this technology needs to be further explored. The paper "Research on Detection Method of Abnormal Data of Power Grid Transmission Based on Parallel Classification Algorithm" (Xiong Xuefeng, Zhang Han, Rong Gongli, et al. [J]. Electronic Design Engineering, 2020, 28(24): 91-94, 99.) extracts abnormal data features of power transmission based on information entropy, collects power grid transmission data using wireless mesh network structure, and detects abnormal data of power grid equipment through parallel classification algorithm, but the detection accuracy still has a certain error. Therefore, how to quickly and accurately realize the comprehensive detection and repair of abnormal data of power grid equipment ledger is a major problem that needs to be urgently solved in power grid asset management. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned problems existing in the prior art and to provide a system and method for automatically repairing abnormal data in the asset ledger of power grid equipment with high accuracy and reliability.

[0005] To achieve the above objectives, the present invention provides the following technical solutions:

[0006] A power grid equipment asset ledger abnormal data automatic repair system, comprising a power grid asset data management module and an abnormal data repair module, wherein the abnormal data repair module comprises an abnormal data positioning marking module, an abnormal data clustering analysis module, and an abnormal data automatic repair module, wherein the input end of the abnormal data positioning marking module is connected to the output end of the power grid asset data management module, and the output end of the abnormal data positioning marking module is connected to the input end of the abnormal data automatic repair module through the abnormal data clustering analysis module;

[0007] The power grid asset data management module is used to obtain the original data information of power grid equipment assets from the external system. The original data information of power grid equipment assets includes basic data of various types of equipment asset ledgers, equipment asset value scale data, equipment asset financial data, basic data of scrapped equipment assets, and historical scrapped and retired equipment data;

[0008] The abnormal data location marking module is used to identify and mark abnormal fields in the original data of power grid equipment assets, and determine the data abnormality type and associated fields corresponding to each abnormal field;

[0009] The abnormal data clustering analysis module is used to build a weighted clustering algorithm model based on the marked abnormal fields, perform clustering analysis on the marked abnormal fields according to the data abnormality type, and form multiple abnormal classification data sets;

[0010] The abnormal data automatic repair module is used to repair the abnormal data according to the abnormal type to which each abnormal classification data set belongs, using a repair solution corresponding to the abnormal type.

[0011] The system also includes a repair result management module, which includes a repaired data output display module and a data storage management module. The input end of the repaired data output display module is connected to the output end of the abnormal data automatic repair module, and the output end of the repaired data output display module is connected to the input end of the data storage management module.

[0012] A method for automatically repairing abnormal data in a power grid equipment asset ledger comprises the following steps in sequence:

[0013] Step A: the power grid asset data management module obtains the original data information of the power grid equipment assets from the external system and sends it to the abnormal data positioning marking module;

[0014] Step B: the abnormal data location marking module identifies and marks abnormal fields in the original data of the power grid equipment assets, and determines the data abnormality type and associated fields corresponding to each abnormal field;

[0015] Step C: The abnormal data clustering analysis module constructs a weighted clustering algorithm model based on the marked abnormal fields, performs clustering analysis on the marked abnormal fields according to the data abnormality type, and forms multiple abnormal classification data sets;

[0016] Step D: The abnormal data automatic repair module repairs the abnormal data using a repair solution corresponding to the abnormal type to which each abnormal classification data set belongs.

[0017] In step C, the exception types include limited enumeration exception class, numerical exception class and time exception class. The limited enumeration exception class includes null value, non-standard value and logical abnormality of value between fields. The numerical exception class includes null value, value less than or equal to 0, too large value and logical abnormality of value between fields. The time exception class includes null value, early or late value and logical abnormality of value between fields.

[0018] In step D, the repair solution corresponding to the abnormality type is:

[0019] For the case where the value in the finite enumeration exception class is empty, or the value in the numerical exception class is empty and has a default value, fix it by setting it to the default value or obtaining the accurate value through the standard comparison table of associated device models;

[0020] For the case where the value in the finite enumeration exception class is non-standard, the solution of obtaining the standard value by associating the standard comparison table is used for repair;

[0021] For the case of logical anomalies in the value selection between fields in the finite enumeration exception class, the solution of synchronous replacement of association between fields is used to repair it;

[0022] For the cases where the value in the numerical anomaly field is empty and there is no default value, and the value in the numerical anomaly class is other than empty, the solution is to perform k-means clustering or data binning algorithm prediction on the same type of assets and the same field value;

[0023] For the case of logical anomalies in the value selection between fields in the time anomaly class, the solution of adjusting the order or associating redundant information fields through multi-source system business data is adopted to repair it;

[0024] For the time anomaly category, except for the logic anomaly of the value between fields, the solution of associating redundant information fields with the business data of multiple source systems is used to repair the problem;

[0025] For fields that are both finite enumeration anomalies and numerical anomalies, when the aforementioned methods cannot be used to repair them, the values ​​of the fields involved are clustered using a weighted clustering algorithm, and the cluster center value with the closest comprehensive distance between the values ​​of multiple fields and any cluster center is replaced and repaired.

[0026] In step B, the abnormal field includes at least one of the equipment asset financial amount, voltage level, quantity scale, technology scale, value scale, equipment model, manufacturer, factory date, commissioning date, and capital transfer date, wherein the voltage level, equipment model, and manufacturer abnormalities are limited enumeration abnormalities, the equipment asset financial amount, quantity scale, technology scale, and value scale abnormalities are numerical abnormalities, and the factory date, commissioning date, and capital transfer date abnormalities are time abnormalities;

[0027] For voltage level abnormal fields, when the aforementioned methods cannot be used to repair them, natural language processing technology is used to extract the voltage level and device name in the system for repair.

[0028] In step B, the method for identifying and marking abnormal fields in the original data of power grid equipment assets is at least one of threshold value checking, binning operation, and standard value matching.

[0029] The system further comprises a repair result management module, which comprises a repaired data output display module and a data storage management module, wherein the signal input end of the repaired data output display module is connected to the signal output end of the abnormal data automatic repair module, and the signal output end of the repaired data output display module is connected to the signal input end of the data storage management module;

[0030] The method further comprises:

[0031] Step E: The repaired data output display module displays the repaired data and compares it with the data before repair to determine whether the repair is successful. If so, the repaired data is output to the data storage management module. If not, return to step B to repair again.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The present invention discloses an automatic repair system for abnormal data of a power grid equipment asset account, including a power grid asset data management module and an abnormal data repair module, wherein the abnormal data repair module includes an abnormal data positioning marking module, an abnormal data clustering analysis module, and an abnormal data automatic repair module, wherein the power grid asset data management module is used to obtain the original data information of power grid equipment assets from an external system, the abnormal data positioning marking module is used to identify and mark the abnormal fields in the original data of power grid equipment assets, and determine the data abnormality type and associated fields corresponding to each abnormal field, the abnormal data clustering analysis module is used to construct a weighted clustering algorithm model based on the marked abnormal fields, cluster analysis is performed on the marked abnormal fields according to the data abnormality type, and multiple abnormal classification data sets are formed, and the abnormal data automatic repair module is used to repair abnormal data according to the abnormal type to which each abnormal classification data set belongs, and adopts the repair scheme corresponding to the abnormal type, and the system adopts different abnormal repair schemes for different abnormal types after identifying and marking the abnormal fields in the data and clustering analysis, thereby not only realizing the rapid positioning detection and automatic repair of abnormal data of power grid equipment asset account, but also having high accuracy and reliability. Therefore, the present invention realizes the rapid, accurate and reliable positioning detection and automatic repair of abnormal data of power grid equipment asset account. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0035] The present invention will be further described below in conjunction with the accompanying drawings and specific implementation methods.

[0036] See also Figure 1 , a power grid equipment asset ledger abnormal data automatic repair system, including a power grid asset data management module 1, an abnormal data repair module 2, the abnormal data repair module 2 includes an abnormal data positioning marking module 21, an abnormal data clustering analysis module 22, and an abnormal data automatic repair module 23, the input end of the abnormal data positioning marking module 21 is connected to the output end of the power grid asset data management module 1, and the output end of the abnormal data positioning marking module 21 is connected to the input end of the abnormal data automatic repair module 23 through the abnormal data clustering analysis module 22;

[0037] The power grid asset data management module 1 is used to obtain the original data information of power grid equipment assets from the external system. The original data information of power grid equipment assets includes basic data of various types of equipment asset ledgers, equipment asset value scale data, equipment asset financial data, basic data of scrapped equipment assets, and historical scrapped and retired equipment data;

[0038] The abnormal data location marking module 21 is used to identify and mark abnormal fields in the original data of power grid equipment assets, and determine the data abnormality type and associated fields corresponding to each abnormal field;

[0039] The abnormal data clustering analysis module 22 is used to build a weighted clustering algorithm model based on the marked abnormal fields, perform clustering analysis on the marked abnormal fields according to the data abnormality type, and form multiple abnormal classification data sets;

[0040] The abnormal data automatic repair module 23 is used to repair the abnormal data by using a repair solution corresponding to the abnormal type to which each abnormal classification data set belongs.

[0041] The system also includes a repair result management module 3, which includes a repaired data output display module 31 and a data storage management module 32. The input end of the repaired data output display module 31 is connected to the output end of the abnormal data automatic repair module 23, and the output end of the repaired data output display module 31 is connected to the input end of the data storage management module 32.

[0042] A method for automatically repairing abnormal data in a power grid equipment asset ledger comprises the following steps in sequence:

[0043] Step A: the power grid asset data management module 1 obtains the original data information of the power grid equipment assets from the external system and sends it to the abnormal data positioning marking module 21;

[0044] Step B: the abnormal data location marking module 21 identifies and marks abnormal fields in the original data of the power grid equipment assets, and determines the data abnormality type and associated fields corresponding to each abnormal field;

[0045] Step C: the abnormal data clustering analysis module 22 constructs a weighted clustering algorithm model based on the marked abnormal fields, performs clustering analysis on the marked abnormal fields according to the data abnormality type, and forms multiple abnormal classification data sets;

[0046] Step D: The abnormal data automatic repair module 23 repairs the abnormal data by using a repair solution corresponding to the abnormal type to which each abnormal classification data set belongs.

[0047] In step C, the exception types include limited enumeration exception class, numerical exception class and time exception class. The limited enumeration exception class includes null value, non-standard value and logical abnormality of value between fields. The numerical exception class includes null value, value less than or equal to 0, too large value and logical abnormality of value between fields. The time exception class includes null value, early or late value and logical abnormality of value between fields.

[0048] In step D, the repair solution corresponding to the abnormality type is:

[0049] For the case where the value in the finite enumeration exception class is empty, or the value in the numerical exception class is empty and has a default value, fix it by setting it to the default value or obtaining the accurate value through the standard comparison table of associated device models;

[0050] For the case where the value in the finite enumeration exception class is non-standard, the solution of obtaining the standard value by associating the standard comparison table is used for repair;

[0051] For the case of logical anomalies in the value selection between fields in the finite enumeration exception class, the solution of synchronous replacement of association between fields is used to repair it;

[0052] For the cases where the value in the numerical anomaly field is empty and there is no default value, and the value in the numerical anomaly class is other than empty, the solution is to perform k-means clustering or data binning algorithm prediction on the same type of assets and the same field value;

[0053] For the case of logical anomalies in the value selection between fields in the time anomaly class, the solution of adjusting the order or associating redundant information fields through multi-source system business data is adopted to repair it;

[0054] For the time anomaly category, except for the logic anomaly of the value between fields, the solution of associating redundant information fields with the business data of multiple source systems is used to repair the problem;

[0055] For fields that are both finite enumeration anomalies and numerical anomalies, when the aforementioned methods cannot be used to repair them, the values ​​of the fields involved are clustered using a weighted clustering algorithm, and the cluster center value with the closest comprehensive distance between the values ​​of multiple fields and any cluster center is replaced and repaired.

[0056] In step B, the abnormal field includes at least one of the equipment asset financial amount, voltage level, quantity scale, technology scale, value scale, equipment model, manufacturer, factory date, commissioning date, and capital transfer date, wherein the voltage level, equipment model, and manufacturer abnormalities are limited enumeration abnormalities, the equipment asset financial amount, quantity scale, technology scale, and value scale abnormalities are numerical abnormalities, and the factory date, commissioning date, and capital transfer date abnormalities are time abnormalities;

[0057] For voltage level abnormal fields, when the aforementioned methods cannot be used to repair them, natural language processing technology is used to extract the voltage level and device name in the system for repair.

[0058] In step B, the method for identifying and marking abnormal fields in the original data of power grid equipment assets is at least one of threshold value checking, binning operation, and standard value matching.

[0059] The system further includes a repair result management module 3, which includes a repaired data output display module 31 and a data storage management module 32. The signal input end of the repaired data output display module 31 is connected to the signal output end of the abnormal data automatic repair module 23, and the signal output end of the repaired data output display module 31 is connected to the signal input end of the data storage management module 32.

[0060] The method further comprises:

[0061] Step E: the repaired data output display module 31 displays the repaired data and compares it with the data before repair to determine whether the repair is successful. If so, the repaired data is output to the data storage management module 32. If not, return to step B to repair again.

[0062] In the present invention, the statistical method for the service life probability distribution of the scale of equipment to be scrapped and scrapped is: based on the collected data set, first count the number of recorded data to calculate the service life, then count the number of recorded data corresponding to each service life, and then calculate the proportion of the number of recorded data corresponding to different service lives in the total amount of recorded data, so as to obtain the probability distribution according to the service life.

[0063] Embodiment 1:

[0064] See also Figure 1 A power grid equipment asset ledger abnormal data automatic repair system includes a power grid asset data management module 1, an abnormal data repair module 2, and a repair result management module 3. The abnormal data repair module 2 includes an abnormal data positioning marking module 21, an abnormal data clustering analysis module 22, and an abnormal data automatic repair module 23. The repair result management module 3 includes a repaired data output display module 31 and a data storage management module 32. The input end of the abnormal data positioning marking module 21 is connected to the output end of the power grid asset data management module 1, and the output end of the abnormal data positioning marking module 21 is connected to the input end of the abnormal data automatic repair module 23 through the abnormal data clustering analysis module 22. The input end of the repaired data output display module 31 is connected to the output end of the abnormal data automatic repair module 23, and the output end of the repaired data output display module 31 is connected to the input end of the data storage management module 32.

[0065] A method for automatically repairing abnormal data in the asset ledger of power grid equipment. The method takes the repair of abnormal data in the basic information of the PMS system ledger and the asset value scale information of the ERP system as the research object, and proceeds in the following steps in sequence:

[0066] 1. The power grid asset data management module 1 obtains the original data information of the power grid equipment assets from the PMS system and the ERP system and sends it to the abnormal data positioning marking module 21;

[0067] 2. The abnormal data positioning and marking module 21 identifies and marks the abnormal fields in the original data of the power grid equipment assets through threshold checking, binning operation, and standard value matching, and determines the data abnormality type and associated fields corresponding to each abnormal field. The abnormal fields identified in this embodiment include the original value of the asset, the net value of the asset, the voltage level, the factory date, the commissioning date, and the transfer date. Among them, the voltage level abnormality is a limited enumeration abnormality class, and there are two types of abnormalities: the value is empty and the value is non-standard. Its auxiliary associated fields are asset equipment model and voltage level abnormal value calibration comparison table. The original value of the asset and the net value of the asset are abnormal values ​​of 0 and the net value of the asset is greater than the original value of the asset. Its auxiliary associated fields include the transfer date and equipment model. The factory date, commissioning date, and transfer date abnormalities are time abnormalities. There are two types of abnormalities: the factory date and the commissioning date are empty, and the time sequence of the three is logically sequential;

[0068] 3. The abnormal data clustering analysis module 22 constructs a weighted clustering algorithm model based on the marked abnormal fields, performs clustering analysis on the marked abnormal fields according to the data abnormality type, and forms multiple abnormal classification data sets;

[0069] 4. The abnormal data automatic repair module 23 repairs the abnormal data using the repair scheme corresponding to the abnormal type of each abnormal classification data set, specifically:

[0070] In the case where the voltage level value is empty or non-standard, first match and associate based on the equipment model, and obtain the accurate value from the equipment model voltage level standard comparison table for repair. If the equipment model does not exist in the equipment model voltage level standard comparison table, natural language processing technology is used to extract similar equipment from the State Grid PMS system / ERP system, and voltage level clustering analysis is performed based on the equipment model to obtain the voltage level with the most consistent features (asset manufacturer, equipment model) for repair;

[0071] For abnormalities in the original value and net value of assets, when there is no abnormality in the original value of assets, the net value calculation and repair is carried out based on the original value of assets and the transfer date according to the depreciation policy, or the net value cluster analysis of assets is carried out in combination with similar assets of the State Grid PMS system / ERP system to obtain a reasonable net value for repair; for the case where the original value and net value of assets are both 0 or the net value of assets is greater than the original value of assets, the net value cluster analysis and prediction of assets based on similar equipment of the PMS system / ERP system is carried out to obtain a reasonable original value and net value for repair;

[0072] In the case where the values ​​of the factory date and the commissioning date are empty, the basic information of the equipment ledger from the State Grid PMS system and the basic information of the asset value from the ERP system of the same equipment are associated through the equipment code to replace and repair the transfer date (the transfer date exists in the ERP system but not in the PMS system);

[0073] In the event that the logical sequence of the production date, operation date and capital transfer date is abnormal, the production date shall be updated based on the capital transfer date or operation date.

[0074] 5. The repaired data output display module 31 displays the repaired data and compares it with the data before repair to determine whether the repair is successful. If so, the repaired data is output to the data storage management module 32. If not, return to step 2 to repair again.

Claims

1. A method for automatically repairing abnormal data in a power grid equipment asset ledger, characterized in that: The method is implemented by an automatic repair system for abnormal data in a power grid equipment asset account, the system comprising a power grid asset data management module (1) and an abnormal data repair module (2), the abnormal data repair module (2) comprising an abnormal data positioning marking module (21), an abnormal data clustering analysis module (22), and an abnormal data automatic repair module (23), the input end of the abnormal data positioning marking module (21) being connected to the output end of the power grid asset data management module (1), and the output end of the abnormal data positioning marking module (21) being connected to the input end of the abnormal data automatic repair module (23) via the abnormal data clustering analysis module (22); The method comprises the following steps in sequence: Step A, the power grid asset data management module (1) obtains original data information of power grid equipment assets from an external system and sends it to the abnormal data positioning marking module (21), wherein the original data information of power grid equipment assets includes basic data of various types of equipment asset ledgers, equipment asset value scale data, equipment asset financial data, basic data of scrapped equipment assets, and historical scrapped and retired equipment data; Step B, the abnormal data positioning and marking module (21) identifies and marks abnormal fields in the original data of the power grid equipment assets, and determines the data abnormality type and associated fields corresponding to each abnormal field; Step C, the abnormal data clustering analysis module (22) constructs a weighted clustering algorithm model based on the marked abnormal fields, performs clustering analysis on the marked abnormal fields according to the data abnormality type, and forms multiple abnormal classification data sets, wherein the abnormality type includes a limited enumeration abnormality class, a numerical abnormality class, and a time abnormality class, wherein the limited enumeration abnormality class includes a value of null, a value of non-standard value, and a logical abnormality of values ​​between fields, wherein the numerical abnormality class includes a value of null, a value of less than or equal to 0, a value of too large, and a logical abnormality of values ​​between fields, and wherein the time abnormality class includes a value of null, a value of too early or too late, and a logical abnormality of values ​​between fields; Step D: The abnormal data automatic repair module (23) repairs the abnormal data using a repair solution corresponding to the abnormal type to which each abnormal classification data set belongs, wherein the repair solution corresponding to the abnormal type is: For the case where the value in the finite enumeration exception class is empty, or the value in the numerical exception class is empty and has a default value, fix it by setting it to the default value or obtaining the accurate value through the standard comparison table of associated device models; For the case where the value in the finite enumeration exception class is non-standard, the solution of obtaining the standard value by associating the standard comparison table is used for repair; For the case of logical anomalies in the value selection between fields in the finite enumeration exception class, the solution of synchronous replacement of association between fields is used to repair it; For the cases where the value in the numerical anomaly field is empty and there is no default value, and the value in the numerical anomaly class is other than empty, the solution is to perform k-means clustering or data binning algorithm prediction on the same type of assets and the same field value; For the case of logical anomalies in the value selection between fields in the time anomaly class, the solution of adjusting the order or associating redundant information fields through multi-source system business data is adopted to repair it; For the time anomaly category, except for the logic anomaly of the value between fields, the solution of associating redundant information fields with the business data of multiple source systems is used to repair the problem; For fields that are both finite enumeration anomalies and numerical anomalies, when the aforementioned methods cannot be used to repair them, the values ​​of the fields involved are clustered using a weighted clustering algorithm, and the cluster center value with the closest comprehensive distance between the values ​​of multiple fields and any cluster center is replaced and repaired.

2. A method for automatically repairing abnormal data in a power grid equipment asset account according to claim 1, characterized in that: In step B, the abnormal field includes at least one of the equipment asset financial amount, voltage level, quantity scale, technical scale, value scale, equipment model, manufacturer, factory date, commissioning date, and transfer date, wherein the voltage level, equipment model, and manufacturer abnormalities are limited enumeration abnormalities, the equipment asset financial amount, quantity scale, technical scale, and value scale abnormalities are numerical abnormalities, and the factory date, commissioning date, and transfer date abnormalities are time abnormalities; For voltage level abnormal fields, when the aforementioned methods cannot be used to repair them, natural language processing technology is used to extract the voltage level and device name in the system for repair.

3. A method for automatically repairing abnormal data in a power grid equipment asset account according to claim 1, characterized in that: In step B, the method for identifying and marking abnormal fields in the original data of power grid equipment assets is at least one of threshold value checking, binning operation, and standard value matching.

4. A method for automatically repairing abnormal data in a power grid equipment asset account according to claim 1, characterized in that: The system further comprises a repair result management module (3), the repair result management module (3) comprising a repaired data output display module (31) and a data storage management module (32), the signal input end of the repaired data output display module (31) being connected to the signal output end of the abnormal data automatic repair module (23), and the signal output end of the repaired data output display module (31) being connected to the signal input end of the data storage management module (32); The method further comprises: Step E: the repaired data output display module (31) displays the repaired data and compares it with the data before repair to determine whether the repair is successful. If so, the repaired data is output to the data storage management module (32). If not, the module returns to step B and performs repair again.

Images