A method for automatically identifying defect information entry and supplementing it through drones
By collecting and cleaning defect information from the power grid management platform, and using semantic recognition and character matching algorithms to automatically identify and supplement the defect information found during drone patrols, the large workload and low efficiency of operation and maintenance personnel caused by the input method of transmission line defect information is solved. The problem of this has improved the quality of entry and work efficiency.
Patent Information
- Application Number
- CN202210455587.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-04-28
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2042-04-28
AI Technical Summary
In the prior art, the defect information input method of transmission line leads to large mechanical workload and low work efficiency of operation and maintenance personnel, and the defect information is misrecorded and misrecorded.
By collecting the defect information total table from the power grid management platform for data cleaning and conversion, using semantic recognition and character matching algorithms to automatically identify uninitial defect information, and supplementally register through drone inspection records to form an entered defect sub-table and judge Enter the situation and trigger the drone re-entry.
Automatic comparison and supplementary defect information are realized, which reduces the mechanical workload of operation and maintenance personnel, improves work efficiency, ensures the quality of input, and solves the problems of large workload and low efficiency of operation and maintenance personnel when processing defect information.
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of power transmission line defect information management, and in particular to a method for automatically identifying defect information entry conditions and supplementing the recording through a drone. Background Art
[0002] Transmission lines are an important part of the power system and bear the heavy responsibility of power transmission. Timely detection of transmission defects and control of the transmission defect processing process are of great significance to the safe operation of the power system. Transmission defects are mainly discovered through daily inspections by operation and maintenance personnel and drone inspections. Operation and maintenance personnel not only need to enter the defects found in daily inspections into the defect management system, but also need to enter the defects that are not found in daily inspections but are found by drone inspections. There will inevitably be duplication of defect information in both methods, and the number of transmission defect information detected by drones is nearly 100,000 throughout the year. There is a lot of data comparison work, which greatly increases the mechanical workload of operation and maintenance personnel and greatly reduces work efficiency. At the same time, there will be situations such as mis-recording and missing defect information. Summary of the invention
[0003] The technical problem to be solved by the present invention is to provide a method for automatically identifying the entry status of defect information and supplementing it through a drone, so as to solve the technical problems that the existing method of entering defect information of power transmission lines will cause a large mechanical workload for operation and maintenance personnel, low work efficiency, and erroneous and missing defect information.
[0004] The technical solution of the present invention is:
[0005] A method for automatically identifying defect information entry and supplementing it through a drone, the method comprising: collecting a defect information general table LrTb_z from a power grid management platform, performing data cleaning and conversion, and extracting key valid fields; creating a set of operating unit abbreviations; using semantic recognition and character matching algorithms to analyze the data to form an entry defect sub-table LrTb_f(x); preliminarily screening n qualified entry defects according to the name of the drone patrol defect route and the tower number, and if n≤1, it is determined that the drone patrol defect has not been entered, triggering the drone patrol record to match the defect information for supplementary recording; if n≥1, calculating the maximum value η1 of the matching rate of "defect appearance + defect description" in the qualified entry defects, if η1≤2 / 3, it is determined that the drone patrol defect has not been entered, triggering the drone patrol record to match the defect information for supplementary recording, and if η1≥2 / 3, it is determined that the drone patrol defect has been entered; judging according to the maximum matching rate η2_max of the "defect description" content, the unentered defect information is entered into the system.
[0006] The method for collecting defect information from the power grid management platform and performing data cleaning and conversion to extract key valid fields is as follows: performing data cleaning and conversion on the defect summary table LrTb_z entered, standardizing the Roman letters in the "location" field, i.e., the line name content, replacing the Roman letters with English letters with the corresponding Roman letters, merging the "functional location", "equipment name" and "defect manifestation" field contents, and deleting the voltage level information.
[0007] The method for generating the input defect sub-table is as follows: obtain the abbreviation of each operating unit in order from the operating unit abbreviation set, and filter the "unit" field of the input defect general table LrTb_z according to the abbreviation, extract the screening results to form the input defect sub-table LrTb_f(x) corresponding to each operating unit, where x is the serial number of each unit; obtain the machine patrol defect record in order from the machine patrol defect general table JxTb_z, and locate the corresponding input defect sub-table LrTb_f(x) according to the "belonging unit" field in the record.
[0008] The method for initially screening the defects that meet the conditions for entry according to the machine patrol defect route name and tower number is: use the "route name" and "tower number" in the machine patrol defect record as screening conditions, and screen the records that meet the conditions in LrTb_f(x). If the number of screening results is 0, it is determined that the machine patrol defect record has not been entered into the defect management system.
[0009] The method for determining the maximum matching rate η1 is as follows: for each qualified input defect record screened out, the "defect appearance + defect description" content is extracted respectively, and the character matching algorithm is used to calculate the matching rate with the "defect appearance + defect description" content of the machine inspection defect record, and the maximum matching rate η1_max is obtained.
[0010] The method for determining the maximum matching rate η2_max of the "defect description" content is as follows: for each qualified input defect record screened out, its "defect description" content is extracted and the character matching algorithm is used to calculate the matching rate with the "defect description" content of the machine patrol defect record to obtain the maximum matching rate η2_max.
[0011] The method for entering the defect information that has not been entered into the system based on the maximum matching rate η2_max of the "defect description" content is as follows: if η2_max is greater than or equal to 1 / 3, it is determined that the drone patrol defect has been entered into the defect management system; otherwise, it is determined that the drone patrol defect has not been entered into the defect management system. If it has not been entered, the drone inspection record is triggered to match the defect information and enter it.
[0012] It also includes: reading the unit sequence, sequentially obtaining the corresponding sub-tables of the unit entry defects, merging the unit entry defect sub-tables into the merged table, determining whether the end of the unit sequence has been reached, and ending the entry.
[0013] Beneficial effects of the present invention:
[0014] The present invention establishes an automatic defect information comparison model and method, and uses semantic recognition and character matching algorithms to automatically identify information that was not found or entered in daily inspections discovered by drone inspections, and supplements the information through drone inspection records; the present invention aims to solve the problems of heavy workload for defect information comparison, low efficiency in defect information entry, and uncontrollable entry quality for power transmission operation and maintenance personnel, and provides a method for automatically identifying a large amount of defect information entry and supplementing the information through drones, which can be used to control the defect entry quality and defect handling process, reduce the mechanical workload of operation and maintenance personnel, improve their work efficiency, provide effective control means for defect management, and improve work quality; and solves the technical problems that the prior art method of power transmission line defect information entry will cause heavy mechanical workload for operation and maintenance personnel, low work efficiency, and erroneous and missing defect information recording. DETAILED DESCRIPTION
[0015] The present invention discloses a method for automatically identifying the input status of defect information and supplementing the input through a drone. The method collects a defect information master table LrTb_z from a power grid management platform, performs data cleaning and conversion, and extracts key valid fields; creates a set of abbreviations of operating units; uses semantic recognition and character matching algorithms to analyze the data and forms an input defect sub-table; preliminarily screens the input defects (the number is n) that meet the conditions according to the name of the drone patrol defect route and the tower number; if n≤1, it is determined that the drone patrol defect has not been input, and the drone patrol record is triggered to match the defect information for supplementary recording; if n≥1, the maximum value η1 of the matching rate of "defect appearance + description" in the input defects that meet the conditions is calculated; if η1≤2 / 3, it is determined that the drone patrol defect has not been input, and the drone patrol record is triggered to match the defect information for supplementary recording; if η1≥2 / 3, it is determined that the drone patrol defect has been input; reads the unit sequence, sequentially obtains the corresponding sub-tables of the unit input defects, merges the unit input defect sub-tables into the merged table, determines whether it has reached the end of the unit sequence, and ends the input; specifically includes:
[0016] Step 1: Collect the defect information summary table LrTb_z from the power grid management platform, clean and convert the data entered into the defect summary table LrTb_z, standardize the Roman letters in the "location" field (i.e., line name), that is, replace the Roman letters with English letters (such as "XII" with "Ⅻ"), merge the "function location", "equipment name", and "defect appearance" fields, and delete the voltage level information, so as to facilitate the subsequent comparison using the character matching algorithm;
[0017] Step 2, create an operation unit sequence to form an operation unit abbreviation set {Guiyang, Zunyi, Liupanshui, Anshun, Kaili, Duyun, Xingyi, Bijie, Tongren, Gui'an, transportation and inspection};
[0018] Step 3, obtain the abbreviation of each operating unit from the operating unit abbreviation set in order, and filter the "unit" field of the input defect general table LrTb_z according to the abbreviation, extract the screening results to form the input defect sub-table LrTb_f(x) corresponding to each operating unit (x is the serial number of each unit), and save it for future use;
[0019] Step 4, obtain the aircraft patrol defect records in order from the aircraft patrol defect summary table JxTb_z, and locate the corresponding entry defect sub-table LrTb_f(x) according to the "belonging unit" field in the record (x is the serial number of each unit);
[0020] Step 5: Use the "line name" (remove information such as voltage level and keep only key information of line naming) and "tower number" (remove "#" and "0" character at the beginning of the name) in the machine inspection defect record as screening conditions to preliminarily screen the records that meet the conditions in LrTb_f(x). If the number of screening results is 0, it is determined that the machine inspection defect record has not been entered into the defect management system, otherwise proceed to step 6 for further comparison;
[0021] Step 6: For each qualified input defect record selected in step 5, extract its "defect appearance + defect description" content, and use the character matching algorithm to calculate the matching rate between it and the "defect appearance + defect description" content of the machine inspection defect record, and obtain the maximum matching rate η1_max;
[0022] Step 7, if η1_max is greater than or equal to 2 / 3, it is determined that the machine inspection defect has been entered into the defect management system, otherwise it proceeds to step 8 for further comparison;
[0023] Step 8: For each qualified input defect record selected in step 5, extract its "defect description" content respectively, and use the character matching algorithm to calculate the matching rate between it and the "defect description" content of the machine inspection defect record, and obtain the maximum matching rate η2_max;
[0024] Step 9: If η2_max is greater than or equal to 1 / 3, it is determined that the drone inspection defect has been entered into the defect management system; otherwise, it is determined that the drone inspection defect has not been entered into the defect management system. If it has not been entered, the drone inspection record is triggered to match the defect information and enter it;
[0025] Step 10, repeating steps 4 to 9 until each record in the machine inspection defect summary table JxTb_z has completed the determination of whether it has been entered into the defect management system;
[0026] Step 11: Summarize and count the results of whether each machine inspection defect has been entered into the defect management system, and obtain indicators such as the number of entries and entry rate of each unit.
Claims
1. A method for automatically identifying defect information entry and supplementing it through drones, Features: The method comprises the following steps: collecting a defect information general table LrTb_z from a power grid management platform, performing data cleaning and conversion, and extracting key valid fields; creating a set of abbreviations of operating units; using semantic recognition and character matching algorithms to analyze the data and form an input defect sub-table LrTb_f(x); preliminarily screening the input defects that meet the conditions according to the names of the defect routes of the patrol machines and the tower numbers, the number of which is n; if n≤1, it is determined that the patrol machine defects have not been input, and the drone inspection records are triggered to match the defect information for supplementary recording; if n≥1, the maximum matching rate η1_max of "defect appearance + defect description" in the input defects that meet the conditions is calculated; if η1_max≤2 / 3, it is determined that the patrol machine defects have not been input, and the drone inspection records are triggered to match the defect information for supplementary recording; if η1_max≥2 / 3, it is determined that the patrol machine defects have not been input, and the drone inspection records are triggered to match the defect information for supplementary recording; if η1_max≥2 / 3, it is determined that the patrol machine defects have been input; According to the maximum matching rate η2_max of the "defect description" content, the defect information that has not been entered is judged to be entered into the system; the method for determining the maximum matching rate η1_max is: for each qualified entry defect record screened out, the "defect appearance + defect description" content is extracted respectively, and the character matching algorithm is used to calculate the matching rate with the "defect appearance + defect description" content of the aircraft patrol defect record, and the maximum matching rate η1_max is obtained; the method for determining the maximum matching rate η2_max of the "defect description" content is: for each qualified entry defect record screened out, its "defect description" content is extracted respectively, and the character matching algorithm is used to calculate the matching rate with the "defect description" content of the aircraft patrol defect record, and the maximum matching rate η2_max is obtained; the method for judging that the defect information that has not been entered into the system according to the maximum matching rate η2_max of the "defect description" content is: if η2_max is greater than or equal to 1 / 3, it is judged that the aircraft patrol defect has been entered into the defect management system, otherwise it is judged that the aircraft patrol defect has not been entered into the defect management system, and if it has not been entered, the drone inspection record is triggered to match the defect information and enter it.
2. According to claim 1, a method for automatically identifying defect information input and supplementing it by drone, Features: The method for collecting defect information from the power grid management platform and performing data cleaning and conversion to extract key valid fields is as follows: performing data cleaning and conversion on the defect table LrTb_z entered, standardizing the Roman letters in the "location" field, i.e., the line name content, replacing the Roman letters with English letters with the corresponding Roman letters, merging the "functional location", "equipment name" and "defect manifestation" field contents, and deleting the voltage level information.
3. According to claim 1, a method for automatically identifying defect information input and supplementing it by drone, Features: The method for generating the input defect sub-table is as follows: obtain the abbreviation of each operating unit in order from the operating unit abbreviation set, and filter the "unit" field of the input defect general table LrTb_z according to the abbreviation, extract the filtering results to form the input defect sub-table LrTb_f(x) corresponding to each operating unit, where x is the serial number of each unit; Obtain the aircraft inspection defect records in order from the aircraft inspection defect summary table JxTb_z, and locate the corresponding entry defect sub-table LrTb_f(x) according to the "belonging unit" field in the record.
4. According to claim 1, a method for automatically identifying defect information input and supplementing it by drone, Features: The method for initially screening the defects that meet the conditions for entry according to the machine patrol defect route name and tower number is: use the "route name" and "tower number" in the machine patrol defect record as screening conditions, and screen the records that meet the conditions in LrTb_f(x). If the number of screening results is 0, it is determined that the machine patrol defect record has not been entered into the defect management system.
5. According to claim 1, a method for automatically identifying defect information input and supplementing it by drone, Features: It also includes: reading the unit sequence, sequentially obtaining the corresponding sub-tables of the unit entry defects, merging the unit entry defect sub-tables into the merged table, determining whether the end of the unit sequence has been reached, and ending the entry.
Citation Information
Patent Citations
Method and device for calculating similarity of text data
CN102214232A
An air-ground integrated transmission line intelligent machine patrol operation system and a method
CN109256723A
Text code combined defect reason automatic classification method
CN109492106A
Cable character detection method and system based on OCR
CN110197181A
Character matching method and device
CN110334711A
