Distribution line conductor defect hidden danger positioning method based on large model and impedance optimization
Through large-scale modeling and impedance optimization technology, the distribution lines are simulated and modeled and data preprocessed, and the conductor defects are automatically analyzed, which solves the problem of increased impedance of distribution lines and improves the operating reliability and efficiency of distribution lines.
Patent Information
- Application Number
- CN202410964650.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2024-11-08
AI Technical Summary
Due to long-term exposure to complex environments and physical corrosion, distribution lines are prone to aging and damage, resulting in increased impedance, which increases operating costs and reduces the reliability and efficiency of power supply, posing a challenge to the stability of the power system.
Using a method based on large model and impedance optimization, the target distribution line is simulated and modeled and data preprocessed, and the semantic understanding ability of the large model is analyzed and processed, and the wire defect hidden danger information is automatically output to achieve line impedance optimization.
It reduces the number of on-site visits by distribution line operation and maintenance personnel, reduces non-essential human resource consumption, and improves the operating reliability and efficiency of distribution line.
Smart Images

Figure CN118916403A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of distribution network operation and maintenance, and in particular to a method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization. Background Art
[0002] Big models are currently cutting-edge application technologies. They simulate human understanding, reasoning, and creativity by learning massive data sets, and are widely used in natural language processing, image recognition, decision support, and other fields.
[0003] The large model can analyze the power grid operation data in real time, predict the change trend of the power grid load, provide scientific dispatching solutions for dispatchers, and ensure the stable operation of the power grid. At the same time, it can also respond quickly to emergencies, provide decision support for emergency repairs, and reduce power outage time and losses.
[0004] The big model can predict the time and type of equipment failure through in-depth mining of equipment operation data, and implement preventive maintenance of equipment. This can not only extend the service life of equipment, but also reduce maintenance costs and improve the economic benefits of power companies.
[0005] The distribution network is a key terminal of the power system, directly supplying electricity to residents and various industries. However, due to long-term exposure to complex environments and physical corrosion, as well as the impact of urban construction, distribution lines are prone to aging and damage, increasing impedance, which not only increases operating costs, but also reduces the reliability and efficiency of power supply, posing a challenge to the stability of the power system.
[0006] Therefore, the present invention optimizes the impedance and utilizes the powerful semantic understanding ability of the cutting-edge technology large model to analyze and process the target distribution line, and automatically outputs the conductor defect hidden danger information of the target distribution line. Summary of the invention
[0007] The present invention proposes an impedance optimization method based on conductor table lookup value impedance fine-tuning. By simulating and modeling the target distribution line, the line impedance is optimized to obtain a refined simulation model of the target distribution line. The powerful semantic understanding ability of the cutting-edge technology large model is used to analyze and process the target distribution line, and the conductor defect hidden danger information of the target distribution line is automatically output.
[0008] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical scheme:
[0009] A method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization, the method comprising the following steps:
[0010] S1, obtaining a first topological structure, first operating data, a report document, and a conductor line resistance lookup table and a reactance lookup table of a target distribution line;
[0011] S2, performing data preprocessing on the first topology structure and the first operation data of the target distribution line to obtain a second topology structure and second operation data;
[0012] S3. Constructing a first simulation model of the target distribution line based on the second topology;
[0013] S4. Based on the second operating data and the first simulation model, line impedance optimization training is performed, and a second simulation model is constructed according to the optimized line impedance;
[0014] S5. Obtain the line impedance of the second simulation model to obtain a line impedance table;
[0015] S6. Use the large model to analyze and process the report document, line impedance table, resistance lookup table and reactance lookup table to obtain the conductor defect hidden danger location information.
[0016] In a further preferred embodiment of the technical solution of the present invention, the first topological structure in step S1 is used to describe the connection relationship between the devices of the target distribution line, the devices of the target distribution line include conductors, substations, switches and distribution transformers, the first topological structure includes parameters and user information of the devices of the target distribution line, and the user information includes user number information;
[0017] The first operation data includes five types of data at multiple time sections of the head end and each node of the target distribution line: current, voltage, active power, reactive power and power factor;
[0018] Each node is a target power distribution line device installed with a measurement device;
[0019] The report documents include fault analysis report, problem analysis report, technical improvement plan report, engineering design report and engineering completion report of the target distribution line.
[0020] In a further preferred embodiment of the technical solution of the present invention, in step S2, data preprocessing is performed on the first topological structure of the target distribution line to obtain a second topological structure, which includes the following steps:
[0021] AS1. Check the length and model information of each wire and whether the user number information of each transformer is missing;
[0022] AS2. For wires with missing length information, the length information is calculated using the longitude and latitude coordinates of the wire's first and last endpoints;
[0023] AS3. For the wire without model information, search for the previous wire. If the previous wire has model information, fill the model information of the previous wire into the wire. If not, continue to search for the previous wire of the previous wire until the model information of the wire is found and filled into the wire.
[0024] AS4. For transformers that are missing user ID information, find the corresponding user in the first operation data according to the transformer name, and fill the user ID information of the user into the transformer.
[0025] In a further preferred embodiment of the technical solution of the present invention, in step S2, preprocessing the first operation data of the target distribution line to obtain the second operation data comprises the following steps:
[0026] BS1. Convert the reading values in the first operation data to the real values;
[0027] BS2. For the five data of the time section of each node, if one of the data is missing, use the missing data for estimation, otherwise execute BS3;
[0028] BS3. Fill in the missing data of the same type at different time sections of each node, as follows:
[0029] The same type of missing data at different time sections of each node are arranged in chronological order, and the proportion of missing values is calculated. If the proportion of missing values is less than the first threshold, linear interpolation is used to fill the missing values. Otherwise, the mean of the non-missing data is used to fill the missing values.
[0030] In a further preferred embodiment of the technical solution of the present invention, the line impedance optimization training in step S4 includes the following steps:
[0031] CS1. Assign subordinate wire groups to each node of the first simulation model;
[0032] CS2. Fill the resistance value and reactance value of each conductor in the target distribution line corresponding to the line type in the resistance lookup table and the reactance lookup table as initial parameters into the first simulation model for simulation calculation;
[0033] CS3. Fill the voltage value at the head end of the target distribution line and the active power and reactive power values of each node into the first simulation model as other initial parameters according to the time section for simulation calculation to obtain multiple voltage calculation values of each node;
[0034] CS4. Statistically calculate multiple voltage calculation values of each node to obtain the average voltage error rate of each node;
[0035] CS5. Determine whether the average voltage error rate of each node is less than the second threshold. If so, keep the current model as the second simulation model and end the step. If not, continue to CS6;
[0036] CS6. For nodes whose average voltage error rate is greater than the second threshold, fine-tune the impedance of their subordinate wire groups and continue with CS3.
[0037] In a further preferred embodiment of the technical solution of the present invention, in step CS6, the impedance is fine-tuned by using an impedance fine-tuning formula, specifically:
[0038] New resistance value = original resistance value + average voltage error rate × resistance fine-tuning coefficient;
[0039] New reactance value = original resistance value + average voltage error rate × reactance fine-tuning coefficient.
[0040] In a further preferred embodiment of the technical solution of the present invention, in step S6, the report document, the line impedance table, the resistance lookup table and the reactance lookup table are analyzed and processed by using a large model to obtain the wire defect hidden danger location information, as follows:
[0041] The large models include the first large model, the second large model and the third large model;
[0042] Input the report document into the first model to obtain the line repair test table;
[0043] The line repair and test table and the line impedance table are input into the second largest model to obtain the second largest model output, and the second largest model output includes the conductor name, the conductor repair and test name, resistance, reactance and the repair and test record corresponding to the conductor;
[0044] The output of the second largest model, the resistance lookup table and the reactance lookup table are input into the third largest model to obtain the wire name and the corresponding evaluation result of the wire, and the defective wire shown in the evaluation result is located according to the corresponding wire name.
[0045] In a further embodiment of the technical solution of the present invention, the method further includes adjusting the simulation parameters of the first simulation model and the second simulation model by using the copper loss of the transformer winding, the iron loss of the transformer core and the transformer transformation ratio.
[0046] A distribution line conductor defect hidden danger location system based on a large model and impedance optimization, the system comprising a data acquisition module, a data processing module, a simulation module and a large model module;
[0047] The data acquisition module is used to acquire the first topological structure, the first operation data, the report document, the resistance lookup table and the reactance lookup table of the target distribution line;
[0048] The data processing module is used to perform data preprocessing on the first topology structure and the first operation data of the target distribution line to obtain the second topology structure and the second operation data;
[0049] The simulation module is used to construct a first simulation model, perform line impedance optimization training according to the first simulation model, and construct a second simulation model according to the optimized line impedance to obtain a line impedance table.
[0050] The large model module is used to analyze and process report documents, line impedance tables, conductor line resistance lookup tables and reactance lookup tables to obtain conductor defect hidden danger location information.
[0051] An electronic device includes a processor and a memory, wherein the memory stores a computer program that can be run on the processor.
[0052] Compared with the prior art, the present invention has the following beneficial effects:
[0053] 1. The first topology data and the first operation data of the present invention can be directly derived from an existing system, and data acquisition is simple.
[0054] 2. The impedance optimization calculation of the present invention is relatively simple and easy to implement.
[0055] 3. The present invention uses the powerful semantic understanding ability of the large model to replace human work and realize the automation of heterogeneous data matching.
[0056] 4. The present invention can obtain the line impedance in the target distribution line, so that the distribution line operation and maintenance personnel can have a certain understanding of the target distribution line in operation at the data level, reducing the number of times the distribution line operation and maintenance personnel repeatedly go to the site and reducing unnecessary human resource consumption. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 It is an overall flow chart of the method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization of the present invention;
[0058] Figure 2 It is a first topology structure preprocessing flow chart of the method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization of the present invention;
[0059] Figure 3 It is a first operation data preprocessing flow chart of the method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization of the present invention;
[0060] Figure 4 It is a line impedance optimization training flow chart of the method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization of the present invention;
[0061] Figure 5 This is an example diagram of a target distribution line according to the method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization of the present invention. DETAILED DESCRIPTION
[0062] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings. It should be understood that the description herein is only used to explain the present invention and is not intended to limit the present invention.
[0063] like Figure 1-5 As shown, a method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization of the present invention is as follows: Figure 1 As shown, the method comprises the following steps:
[0064] S1, obtaining a first topological structure, first operating data, a report document, and a resistance lookup table and a reactance lookup table of corresponding line types of conductors in the first topological structure of a target distribution line;
[0065] S2, performing data preprocessing on the first topology structure and the first operation data of the target distribution line to obtain a second topology structure and second operation data;
[0066] S3. Constructing a first simulation model of the target distribution line based on the second topology;
[0067] S4. Based on the second operating data and the first simulation model, line impedance optimization training is performed, and a second simulation model is constructed according to the optimized line impedance;
[0068] S5. Obtain the line impedance of the second simulation model to obtain a line impedance table;
[0069] S6. Use the large model to analyze and process the report document, line impedance table, resistance lookup table and reactance lookup table to obtain the conductor defect hidden danger location information.
[0070] The first topological structure in step S1 is used to describe the connection relationship between the various devices of the target distribution line. The various devices of the target distribution line include substations, conductors, switches, and distribution transformers; the conductors include overhead conductors and cables, and the switches include load switches, knife switches, circuit breakers, and other devices used to disconnect distribution lines.
[0071] The carrier of the first topology structure is generally an image file or a text file; the image file is generally called a line single-line diagram, and common file formats include PDF, SVG, JPG or PNG; the text file is generally called a line CIM file, and a common file format is XML.
[0072] The first topology structure includes parameters of each device of the target distribution line and user information, and the user information includes user number information.
[0073] The first operation data in step S1 includes the current, voltage, active power, reactive power and power factor of the head end and each node of the target distribution line for a total of 96 time sections in one day; that is, it is recorded every 15 minutes starting from 00:00.
[0074] Each node is a target distribution line device installed with a measuring device, and most of them are automatic switches and distribution transformers installed with a measuring device.
[0075] The report documents in step S1 include a fault analysis report, a problem analysis report, a technical improvement plan report, an engineering design report, a project completion report, etc. of the target distribution line.
[0076] The report document is in text format.
[0077] like Figure 2 As shown, in step S2, preprocessing the first topology structure of the target distribution line to obtain a second topology structure includes the following steps:
[0078] AS1. Check the length and model information of each wire and whether the user number information of each transformer is missing;
[0079] AS2. For wires with missing length information, the length information is calculated using the longitude and latitude coordinates of the wire's first and last endpoints;
[0080] AS3. For the wire without model information, search for the previous wire. If the previous wire has model information, fill the model information of the previous wire into the wire. If not, continue to search for the previous wire of the previous wire until the model information of the wire is found and filled into the wire.
[0081] AS4. For transformers that are missing user numbers, find the corresponding user in the first operation data according to the transformer name, and fill the user number of the user into the transformer.
[0082] For example, in the first topology structure, the name of a transformer with a missing user number is "Hebian Public Transformer". In the first operation data, a user's user name is "Hebian Public Transformer". The two names are different but have been manually confirmed to belong to the same transformer. The user name of "Hebian Public Transformer" in the first operation data can be filled in the "Hebian Public Transformer" transformer in the first topology structure.
[0083] like Figure 3 As shown, in step S2, data preprocessing is performed on the first operation data of the target distribution line to obtain second operation data, including the following steps:
[0084] BS1. Convert the reading values in the first operation data to the real values;
[0085] BS2. For the five data of the time section of each node, if one of the data is missing, use the missing data for estimation, otherwise execute BS3;
[0086] BS3. Fill in the missing data of the same type in different time sections of each node;
[0087] Step BS1 is converted into a real value by the following calculation formula:
[0088] The actual value of current = current reading value × CT ratio. The CT ratio is the ratio of the current on the primary side to the secondary side of the current transformer.
[0089] The actual value of voltage = voltage reading value × PT ratio, where PT is the ratio of the primary side voltage to the secondary side voltage of the voltage transformer;
[0090] The actual value of active power = active power reading value × CT ratio × PT ratio;
[0091] The actual value of five-phase reactive power = reactive power reading value × CT ratio × PT ratio.
[0092] Step BS2 fills in the missing data of the same type in different time sections of each node; the details are as follows:
[0093] The five data of the simultaneous cross section of each node are current, voltage, active power, reactive power and power factor.
[0094] Among the five data of current, voltage, active power, reactive power and power factor, voltage, active power and reactive power are necessary data for simulation calculation. They can be estimated from each other through formulas according to the missing data of the five data in the same time section. The formula is as follows:
[0095] Active power = current × voltage × power factor,
[0096]
[0097] That is, if active power or reactive power is missing, but current, voltage, and power factor are known, the corresponding active power or reactive power value can be estimated by calculation through the formula; if voltage is missing, but current, active power, power factor or current, reactive power, and power factor are known, the corresponding voltage can be estimated by calculation through the formula; if current or power factor is missing, but active power, reactive power, and voltage are known, the corresponding current or power factor can be estimated by calculation through the formula.
[0098] Step BS3 fills in the missing data of the same type in different time sections of each node, as follows:
[0099] Data of the same data type in each node, such as current data of all time sections of a node, voltage data of all time sections of a certain electromechanical device, etc.
[0100] The missing data of the same data type at different time sections of each node are arranged in chronological order, and the missing value ratio is calculated. If the missing value ratio is less than the first threshold, linear interpolation is used to fill the missing value. Otherwise, the mean of the non-missing data is used to fill the missing value. The missing value ratio calculation formula is:
[0101]
[0102] The total number of time points refers to the number of time sections included in the first operation data, and the first threshold is set to 10%.
[0103] The first simulation model and the second simulation model in step S3 and step S4 may be target distribution line simulation models constructed by respective simulation software.
[0104] like Figure 4 As shown, the line impedance optimization training in step S4 specifically includes the following steps:
[0105] CS1. Assign subordinate wire groups to each node of the first simulation model;
[0106] The allocation of subordinate wire groups in step CS1 can be performed in two allocation methods, which can optimize the impedance parameters of the wires according to the simulation results of each node. The method of allocating subordinate wire groups can be more targeted to the impedance parameters that need to be fine-tuned.
[0107] Method 1: All wires on the electrical conduction path from the line head end to the node are subordinate wire groups of the node. In this method, subordinate wire groups of different nodes may overlap.
[0108] Mode 2: Expand the topological structure of the first simulation model into a tree structure with the line head end as the root node. Except for the root node, the remaining tree nodes of the tree structure correspond to the nodes or branch nodes of the first simulation model.
[0109] In the first allocation, all wires between any two tree nodes in the tree structure are subordinate wire groups of the latter tree node; in the second allocation, if the tree node corresponding to the branch node has a subordinate wire group, it is allocated to the next tree node closest to the tree node. If the distance is equal, it is allocated to the next tree node with a shorter branch. After several times of second allocation, all tree nodes corresponding to the branch node have no subordinate wire groups. In this way, the subordinate wire groups of different nodes are independent of each other.
[0110] like Figure 5As shown in the figure, the nodes of the target distribution line are the automatic switches K1 and K2, and the distribution transformers T1, T2, T3, T4, and T5, a total of 7 nodes. The branch nodes are: F1, F2, F3, F4, F5, F6, F7, and F8. It is assumed that the adjacent spacing between the line head end, the automatic switch, and the branch node is 50 meters, and the adjacent spacing between the branch node and the distribution transformer is 10 meters.
[0111] Method 1:
[0112] All the wires on the electrical conduction path from the line head end to the node are subordinate wire groups of the node. In this way, subordinate wire groups of different nodes may have overlapping parts.
[0113] In this way, the subordinate wire groups of each node are:
[0114] Automation switch K1: {[line head end-K1]};
[0115] Automatic switch K2: {[line head end-K1, K1-F1, F1-F2, F2-F3, F3-K2]};
[0116] Distribution transformer T1: {[line head end-K1, K1-F1, F1-F5, F5-T1]};
[0117] Distribution transformer T2: {[line head end-K1, K1-F1, F1-F5, F5-F6, F6-T2]};
[0118] Distribution transformer T3: {[line head end-K1, K1-F1, F1-F2, F2-T3]};
[0119] Distribution transformer T4: {[line head end-K1, K1-F1, F1-F2, F2-F3, F3-K2, K2-F7, F7-F8; F8-T4]};
[0120] Distribution transformer T5: {[line head end-K1, K1-F1, F1-F2, F2-F3, F3-F4, F4-T5]};
[0121] Method 2:
[0122] The topological structure of the first simulation model is expanded into a tree structure with the line head end as the root node. Except for the root node, the remaining tree nodes of the tree structure correspond to the nodes or branch nodes of the first simulation model. In the first allocation, all the wires between any two tree nodes in the tree structure are subordinate wire groups of the latter tree node; in the second allocation, if there is a subordinate wire group, the tree node corresponding to the branch node is allocated to the next tree node closest to the tree node. If the distance is equal, it is allocated to the next tree node with a shorter branch. After several times of second allocation, all tree nodes corresponding to the branch node have no subordinate wire groups.
[0123] In this way, the subordinate wire groups of different nodes are independent of each other.
[0124] In this way, the subordinate wire groups of each node are:
[0125] Automation switch K1: {[line head end-K1]};
[0126] Automation switch K2: {[F2-F3, F3-K2]};
[0127] Distribution transformer T1: {[K1-F1, F1-F5, F5-T1]};
[0128] Distribution transformer T2: {[F5-F6, F6-T2]};
[0129] Distribution transformer T3: {[F1-F2, F2-T3]};
[0130] Distribution transformer T4: {[K2-F7, F7-F8, F8-T4]};
[0131] Distribution transformer T5: {[F3-F4, F4-T5]}.
[0132] CS2 uses the resistance value and reactance value of each conductor in the target distribution line corresponding to the line type in the resistance lookup table and the reactance lookup table as initial parameters to fill in the first simulation model for simulation calculation;
[0133] CS3. Fill the voltage value at the head end of the target distribution line and the active power and reactive power values of each node into the first simulation model as other initial parameters according to the time section for simulation calculation to obtain multiple voltage calculation values of each node;
[0134] CS4. Statistically calculate the multiple voltage calculation values of each node to obtain the average voltage error rate of each node, as follows:
[0135] The voltage error rate is obtained by subtracting the voltage value of each node from the voltage value, that is:
[0136]
[0137] The average voltage error rate is:
[0138]
[0139] m is the number of nodes.
[0140] CS5. Determine whether the average voltage error rate of each node is less than the second threshold. If so, keep the current model as the second simulation model and end the step. If not, continue to CS6;
[0141] The second threshold is generally set to 1%.
[0142] CS6. For nodes whose average voltage error rate is greater than the second threshold, fine-tune the impedance of their subordinate wire groups and continue with CS3.
[0143] The impedance fine-tuning formula is:
[0144] New resistance value = original resistance value + average voltage error rate × resistance fine-tuning coefficient;
[0145] New reactance value = original resistance value + average voltage error rate × reactance fine-tuning coefficient;
[0146] The resistance fine-tuning coefficient and the reactance fine-tuning coefficient are constant coefficients, and their value range is between 0 and 1.
[0147] The method further includes adjusting simulation parameters of the first simulation model and the second simulation model by using the copper loss of the transformer winding, the iron loss of the transformer core and the transformer transformation ratio.
[0148] The line impedance table in step S5 is composed of the resistance value and reactance value of each conductor in the second simulation model.
[0149] The line impedance table includes the conductor name, the resistance value and the reactance value corresponding to the conductor name.
[0150] In step S6, the large model is used to analyze and process the report document, the line impedance table, the wire line resistance lookup table and the reactance lookup table to obtain the wire defect hidden danger location information, specifically:
[0151] The large models include the first large model, the second large model and the third large model;
[0152] Fill the text of the report document into the first prompt, submit the first prompt to the first large model and obtain the output of the first large model to obtain the line history repair and test table;
[0153] The number of texts that can be filled in at one time varies according to the capabilities of the first large model. Therefore, multiple first prompt submissions may be performed to obtain multiple first large model outputs. The line history maintenance test tables output by multiple first large models are collated and merged to obtain the line maintenance test table.
[0154] The line repair and test table includes the name of the repair and test conductor and the corresponding repair and test record.
[0155] Fill the line impedance table and line repair test table into the second prompt, submit the second prompt to the second largest model, and obtain the second largest model output. The second largest model matches the conductor name in the line impedance table and the repair test conductor name in the line repair test table with the conductor name and the repair test conductor name of the same section of conductor, and the conductor name in the line impedance table shall prevail.
[0156] The second largest model output includes the conductor name, repair test conductor name, resistance, reactance and the corresponding repair test record of the conductor.
[0157] Find the wire type of the wire in the second topology structure according to the wire name, find the resistance lookup value and reactance lookup value corresponding to the wire type according to the wire type,
[0158] Fill the third prompt with the output of the second largest model, the line type, resistance table value and reactance table value corresponding to the wire, and submit the third prompt to the third largest model to obtain the output of the third largest model.
[0159] The third model is evaluated based on the conductor name, the conductor line type corresponding to the conductor name, the typical resistance value of the conductor line type which is the resistance lookup value of the conductor line type corresponding to the conductor line type in the conductor line type resistance lookup table and the reactance lookup table, the typical reactance value which is the reactance lookup value of the conductor line type corresponding to the conductor line type, the resistance and reactance values corresponding to the conductor obtained by simulation, and the repair and test records.
[0160] The output of the third largest model includes the conductor name and its corresponding evaluation result. The specific position of the conductor shown as defective in the evaluation result of the third largest model output can be located according to the conductor name. Multiple third largest model outputs are sorted and merged to obtain the location information of all defective conductors in the target distribution line.
[0161] A distribution line conductor defect hidden danger location system based on a large model and impedance optimization, the system comprising a data acquisition module, a data processing module, a simulation module and a large model module;
[0162] The data acquisition module is used to acquire the first topological structure, the first operation data, the report document, the resistance lookup table and the reactance lookup table of the target distribution line;
[0163] The data processing module is used to perform data preprocessing on the first topology structure and the first operation data of the target distribution line to obtain the second topology structure and the second operation data;
[0164] The simulation module is used to construct a first simulation model, perform line impedance optimization training according to the first simulation model, and construct a second simulation model according to the optimized line impedance to obtain a line impedance table.
[0165] The large model module is used to analyze and process report documents, line impedance tables, conductor line resistance lookup tables and line reactance lookup tables to obtain conductor defect hidden danger location information.
[0166] An electronic device includes a processor and a memory, wherein the memory stores a computer program that can be run on the processor.
[0167] The parts not involved in the present invention are the same as the prior art or can be implemented by using the prior art.
[0168] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes in form and details may be made without departing from the spirit and scope of the present invention as defined in the appended claims.
Claims
1. A method for locating defects and hidden dangers of distribution line conductors based on a large model and impedance optimization, characterized in that: The method comprises the following steps: S1, obtaining a first topological structure, first operating data, a report document, and a conductor line resistance lookup table and a reactance lookup table of a target distribution line; S2, performing data preprocessing on the first topology structure and the first operation data of the target distribution line to obtain a second topology structure and second operation data; S3. Constructing a first simulation model of the target distribution line based on the second topology; S4. Based on the second operating data and the first simulation model, line impedance optimization training is performed, and a second simulation model is constructed according to the optimized line impedance; S5. Obtain the line impedance of the second simulation model to obtain a line impedance table; S6. Use the large model to analyze and process the report document, line impedance table, conductor line resistance lookup table and reactance lookup table to obtain the conductor defect hidden danger location information.
2. The method for locating defects and hidden dangers of distribution line conductors based on large models and impedance optimization according to claim 1 is characterized in that: The first topology structure in step S1 is used to describe the connection relationship between the devices of the target distribution line. The devices of the target distribution line include conductors, substations, switches and distribution transformers. The first topology structure includes parameters and user information of the devices of the target distribution line. The user information includes user number information. The first operation data includes five types of data at multiple time sections of the head end and each node of the target distribution line: current, voltage, active power, reactive power and power factor; Each node is a target power distribution line device installed with a measurement device; The report documents include fault analysis report, problem analysis report, technical improvement plan report, engineering design report and engineering completion report of the target distribution line.
3. The method for locating defects and hidden dangers of distribution line conductors based on large models and impedance optimization according to claim 2 is characterized in that: In step S2, data preprocessing is performed on the first topology structure of the target distribution line to obtain a second topology structure, which includes the following steps: AS1. Check the length and model information of each wire and whether the user number information of each transformer is missing; AS2. For wires with missing length information, the length information is calculated using the longitude and latitude coordinates of the wire's first and last endpoints; AS3. For the wire without model information, search for the previous wire. If the previous wire has model information, fill the model information of the previous wire into the wire. If not, continue to search for the previous wire of the previous wire until the model information of the wire is found and filled into the wire. AS4. For transformers that are missing user ID information, find the corresponding user in the first operation data according to the transformer name, and fill the user ID information of the user into the transformer.
4. The method for locating defects and hidden dangers of distribution line conductors based on large models and impedance optimization according to claim 3 is characterized in that: In step S2, data preprocessing is performed on the first operation data of the target distribution line to obtain second operation data, which includes the following steps: BS1. Convert the reading values in the first operation data to the real values; BS2. For the five data of the time section of each node, if one of the data is missing, use the missing data for estimation, otherwise execute BS3; BS3. Fill in the missing data of the same type at different time sections of each node, as follows: The same type of missing data at different time sections of each node are arranged in chronological order, and the proportion of missing values is calculated. If the proportion of missing values is less than the first threshold, linear interpolation is used to fill the missing values. Otherwise, the mean of the non-missing data is used to fill the missing values.
5. The method for locating defects and hidden dangers of distribution line conductors based on large models and impedance optimization according to claim 4 is characterized in that: The line impedance optimization training in step S4 includes the following steps: CS1. Assign subordinate wire groups to each node of the first simulation model; CS2. Fill the resistance value and reactance value of each conductor in the target distribution line corresponding to the line type in the resistance lookup table and the reactance lookup table as initial parameters into the first simulation model for simulation calculation; CS3. Fill the voltage value at the head end of the target distribution line and the active power and reactive power values of each node into the first simulation model as other initial parameters according to the time section for simulation calculation to obtain multiple voltage calculation values of each node; CS4. Statistically calculate multiple voltage calculation values of each node to obtain the average voltage error rate of each node; CS5. Determine whether the average voltage error rate of each node is less than the second threshold. If so, keep the current model as the second simulation model and end the step. If not, continue to CS6; CS6. For nodes whose average voltage error rate is greater than the second threshold, fine-tune the impedance of their subordinate wire groups and continue with CS3.
6. The method for locating defects and hidden dangers of distribution line conductors based on large models and impedance optimization according to claim 5 is characterized in that: In step CS6, the impedance is fine-tuned using the impedance fine-tuning formula, specifically: New resistance value = original resistance value + average voltage error rate × resistance fine-tuning coefficient; New reactance value = original resistance value + average voltage error rate × reactance fine-tuning coefficient.
7. The method for locating defects and hidden dangers of distribution line conductors based on large models and impedance optimization according to claim 6 is characterized in that: In step S6, the report document, line impedance table, resistance lookup table and reactance lookup table are analyzed and processed by the large model to obtain the wire defect hidden danger location information, as follows: The large models include the first large model, the second large model and the third large model; Input the report document into the first model to obtain the line repair test table; The line repair and test table and the line impedance table are input into the second largest model to obtain the second largest model output, and the second largest model output includes the conductor name, the conductor repair and test name, resistance, reactance and the repair and test record corresponding to the conductor; The output of the second largest model, the resistance lookup table and the reactance lookup table are input into the third largest model to obtain the wire name and the corresponding evaluation result of the wire, and the defective wire shown in the evaluation result is located according to the corresponding wire name.
8. The method for locating defects and hidden dangers of distribution line conductors based on large models and impedance optimization according to claim 4 is characterized in that: The method further includes adjusting simulation parameters of the first simulation model and the second simulation model by using the copper loss of the transformer winding, the iron loss of the transformer core and the transformer transformation ratio.
9. A distribution line conductor defect location system based on a large model and impedance optimization, characterized by: The system includes a data acquisition module, a data processing module, a simulation module and a large model module; The data acquisition module is used to acquire the first topological structure, the first operation data, the report document, the resistance lookup table and the reactance lookup table of the target distribution line; The data processing module is used to perform data preprocessing on the first topology structure and the first operation data of the target distribution line to obtain the second topology structure and the second operation data; The simulation module is used to construct a first simulation model, perform line impedance optimization training according to the first simulation model, and construct a second simulation model according to the optimized line impedance to obtain a line impedance table. The large model module is used to analyze and process report documents, line impedance tables, linear resistance lookup tables and linear reactance lookup tables to obtain conductor defect hidden danger location information.
10. An electronic device comprising a processor and a memory, wherein the memory stores a computer program that can be run on the processor, wherein: When the processor runs the computer program, the steps of the method according to any one of claims 1 to 8 are performed.
