A method for line fault diagnosis based on satellite and meteorological information

Abstract

The present invention belongs to the technical field of power fault diagnosis, and in particular to a line fault diagnosis method based on satellites and meteorology, including: regularly collecting satellite images and incorporating them into an image database, identifying risk source information and incorporating them into a hidden danger library; starting a pre-disaster satellite emergency inspection process in the disaster area, incorporating the collection results into the image database, identifying risk source information, calling the data in the hidden danger library for comparison and outputting a list of risk sources in the disaster area; starting a post-fault emergency inspection process in the fault area, incorporating the collection results into the image database, identifying risk source information, calling the data in the hidden danger library for comparison, outputting risk source change information before and after the fault according to the comparison results, and combining the risk source change information, meteorological data at the time of the fault, and fault characteristics at the fault site to screen the source of the fault. In the present invention, by utilizing the real-time data provided by satellites and meteorology, combined with intelligent analysis technology, rapid investigation of foreign body sources in line channels before the disaster and accurate tracing after the fault are achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of power fault diagnosis, and in particular relates to a line fault diagnosis method based on satellite and meteorology. Background Art

[0002] In the current power system, there are many potential risk sources such as industrial plants, agricultural greenhouses, and landscaping along the common line channels. The number of control is large and the difficulty is great. Their operation is seriously affected by natural meteorological factors and nearby human activity risk factors. At present, the inspection methods for the above-mentioned line channels mainly include drone inspections, line monitoring camera inspections, and manual inspections. However, due to the following shortcomings, it is difficult to achieve the goal of accurate and extensive pre-disaster inspections. The specific shortcomings are as follows: Although drone inspections have high accuracy, they are subject to flight approval constraints, high large-scale normalized operation costs, and small monitoring ranges. They cannot meet the needs of normalized, large-scale line channel risk surveys; and line monitoring cameras are limited by monitoring angles and ranges, and cannot fully cover the scope of risk sources; at the same time, if manual inspections are used, labor costs or time costs will increase.

[0003] In addition, the existing methods still have the problem of difficulty in tracing the source of line faults. Although they can rely on drones or manual inspections, they are restricted by weather or environment and cannot find the source of the fault in time and take effective measures. At present, the commonly used methods for fault diagnosis include model-based diagnosis, signal processing technology, and artificial intelligence algorithms. Although these methods can identify and locate faults to a certain extent, they each have their limitations. For example, model-based methods rely on accurate system models, but in actual operation, models are often difficult to fully match the actual situation. Signal processing technology is limited by the quality of signal acquisition and the complexity of analysis algorithms. Although artificial intelligence has shown great potential in pattern recognition, its adaptability to new fault types needs to be improved.

[0004] Based on the above problems, this field needs to explore a low-cost, high-precision, and wide-coverage inspection and fault tracing method, so as to achieve accurate early warning of foreign body hidden dangers before failure, timely and accurate diagnosis after failure, and accurately determine the source of the fault based on the fault diagnosis result information. Summary of the invention

[0005] The present invention provides a line fault diagnosis method based on satellite and meteorology, which can effectively solve the problems in the background technology.

[0006] To achieve the above object, the technical solution of the present invention is as follows:

[0007] A line fault diagnosis method based on satellite and meteorology, comprising:

[0008] Regularly collect satellite images through remote sensing satellites, and incorporate the regular collection results into the image database, wherein the regular collection is carried out in combination with the power grid line channel control requirements;

[0009] Identify the risk source information in the periodic collection results, and include the risk source identification results in the hidden danger database;

[0010] Initiate a pre-disaster satellite emergency patrol process in the disaster area, collect satellite images of the disaster area to obtain a first emergency collection result, and incorporate the first emergency collection result into the image database, wherein the initiation is based on meteorological warning information;

[0011] Identify the risk source information in the first emergency collection result, call the data in the hidden danger database for comparison, and output a list of risk sources in the disaster area according to the comparison result, wherein the list of risk sources in the disaster area is used to guide the disposal of foreign objects at the channel site;

[0012] Initiate a post-fault emergency inspection process in the fault area, collect satellite images of the fault area to obtain a second emergency collection result, incorporate the second emergency collection result into the image database, and initiate based on line channel fault diagnosis information; identify risk source information in the second emergency collection result, and call the data in the hidden danger library for comparison, output risk source change information before and after the fault based on the comparison result, and screen the source of the fault in combination with the risk source change information, meteorological data at the time of the fault, and fault characteristics at the fault site.

[0013] Further, identifying the risk source information in each of the periodic collection results and incorporating the risk source identification results into the hidden danger database includes:

[0014] Identify the risk source information in the satellite images collected in the first phase and include it in the first hidden danger database as a basis;

[0015] Based on the above foundation, a multi-period image comparison is performed on the satellite images collected later to obtain information on changes in risk sources and include the information in the second hidden danger database;

[0016] The first hidden danger database and the second hidden danger database are combined by using an information matching method to obtain the hidden danger database.

[0017] Furthermore, based on the above foundation, a multi-period image comparison is performed on the satellite images collected later to obtain information on changes in risk sources, including:

[0018] The satellite images of two adjacent time phases to be compared are obtained, and the following operations are performed on the satellite images of each time phase:

[0019] Use the target detection model to obtain the extraction results of each target, and use the semantic segmentation model to obtain the extraction results of each target by face, and perform vector processing on the two extraction results of each phase;

[0020] For different phases, risk source selection is performed based on the vector processing result;

[0021] The risk source selection based on two phases obtains the information of risk source changes.

[0022] Furthermore, vector processing is performed on the two extraction results of each phase, including:

[0023] Converting the bounding box extracted by the object detection model into a vector polygon;

[0024] Converting pixel classification results of the semantic segmentation model into vector polygons;

[0025] Perform geometric correction on the vector polygons of target detection and semantic segmentation respectively to achieve alignment with the geographic coordinates of satellite images;

[0026] In the overlapping area of ​​the two vector polygons, the scope of the risk source is defined using the bounding box of the target detection, and the pixel distribution result of the semantic segmentation is used to supplement the area missed by the target detection or with blurred boundaries to complete the merging;

[0027] The merged vector results are used to generate a complete vector dataset, which contains the shape, center point coordinates, location, boundary and spatial attribute information of each risk source.

[0028] Furthermore, the information matching method includes:

[0029] Extract the center point coordinates of the risk sources in the first hidden danger database and the second hidden danger database respectively, set a spatial distance threshold, and compare the center point coordinates of each risk source in the two phases. If the distance between the center point coordinates of the two risk sources is less than the spatial distance threshold, mark the same risk source, otherwise mark it as a new or disappeared risk source;

[0030] Extract boundary polygons for the risk sources marked as the same risk source, calculate the boundary similarity of the risk sources in two phases by shape similarity algorithm, set boundary similarity threshold, if the boundary similarity is higher than the boundary similarity threshold, mark them as the same risk source, otherwise mark them as changed risk sources;

[0031] For the newly added or disappeared risk sources and the changed risk sources, attribute information is extracted and the impact of attribute changes is evaluated, and an impact degree threshold is set. If the impact degree of the attribute change is less than the impact degree threshold, it is marked as the same risk source, otherwise the original marking status is maintained.

[0032] Further, the risk source information in the first emergency collection result is identified, and the data in the hidden danger database is called for comparison, and a list of risk sources in the disaster area is output according to the comparison result, including:

[0033] Based on the comparison with the data in the hidden danger database, the risk source information in the first emergency collection result is classified into a first part that matches the data in the hidden danger database and a second part that does not match;

[0034] Output the list of disaster area risk sources based on the first part.

[0035] Furthermore, the second part is included in the hidden danger library.

[0036] Furthermore, the same risk source information identification method is used for the regular collection results, the first emergency collection results and the second emergency collection results.

[0037] Furthermore, the acquisition of the line channel fault diagnosis information includes:

[0038] Determine the time when the line fails, and start collecting analysis and judgment factor information at the time, wherein the analysis and judgment factor information includes at least line data, online monitoring data, meteorological data and the hidden danger database data; output fault diagnosis information through an intelligent diagnosis model based on the analysis and judgment factor information.

[0039] Furthermore, the intelligent diagnosis model is a multimodal fusion model.

[0040] Furthermore, the source of the fault is screened by combining the risk source change information, the meteorological data at the time of the fault, and the fault characteristics at the fault site, including:

[0041] Based on the risk source change information, identify the change type of the risk source, and determine the impact of each type of change on the line;

[0042] Revising the impact levels based on the meteorological data at the time of the fault to obtain revised results;

[0043] Based on the revision result and the risk source change information, analyzing the matching degree between the fault characteristics and the risk source, and screening the fault source list;

[0044] The screened list of fault sources is prioritized, and the fault source is determined based on the ranking result.

[0045] Furthermore, the pre-disaster satellite emergency inspection process of the disaster area and the post-fault emergency inspection of the fault area both use a portable ground station, which is used to upload a single instruction for image acquisition in the power field and receive the satellite image;

[0046] The process of regularly collecting satellite images adopts a fixed ground station, which is used to centrally upload image collection instructions in several fields, receive image collection data, and transmit the image collection data to a data center.

[0047] The technical solution of the present invention can achieve the following technical effects:

[0048] The present invention utilizes real-time data provided by satellites and meteorology, combined with intelligent analysis technology, to achieve rapid identification of foreign object sources in line channels before disasters and accurate tracing of sources after failures, making the prediction of potential risks of lines more accurate and comprehensive. Operation and maintenance personnel can take corresponding measures quickly in advance, greatly reducing the line failure rate caused by uncontrollable factors such as natural disasters, while improving the accuracy of fault analysis and reducing fault handling time, thereby ensuring the stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 is a flow chart of a line fault diagnosis method based on satellite and meteorology;

[0051] Figure 2 A flowchart for identifying risk source information in the periodic collection results and incorporating the risk source identification results into the hidden danger database;

[0052] Figure 3 A flowchart for comparing multiple periods of satellite images collected later based on the foundation to obtain information on changes in risk sources;

[0053] Figure 4 is a flow chart for performing vector processing on two extraction results of each phase;

[0054] Figure 5 is a flow chart of the information matching method;

[0055] Figure 6 Flow chart for fault source screening;

[0056] Figure 7 This is the flow chart for routine inspections and emergency inspections. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0059] like Figure 1 As shown, a line fault diagnosis method based on satellite and meteorology includes:

[0060] A1: Satellite images are collected regularly through remote sensing satellites, and the regular collection results are included in the image database. Regular collection is carried out in combination with the requirements for power grid line channel control;

[0061] A2: Identify the risk source information in the periodic collection results, and include the risk source identification results in the hidden danger database. The line operation and maintenance personnel will then track and deal with the channel risk sources based on the hidden danger database;

[0062] B1: Start the pre-disaster satellite emergency inspection process in the disaster area, collect satellite images of the disaster area to obtain the first emergency collection results, incorporate the first emergency collection results into the image database, and start based on meteorological warning information, where the sources of meteorological warning information include but are not limited to meteorological stations and power grid meteorological business middle stations;

[0063] B2: Identify the risk source information in the first emergency collection results, call the data in the hidden danger database for comparison, and output the disaster area risk source list based on the comparison results. The disaster area risk source list is used to guide the disposal of foreign objects at the channel site;

[0064] C1: Start the post-fault emergency inspection process in the fault area, collect satellite images of the fault area to obtain the second emergency collection results, incorporate the second emergency collection results into the image database, and start the line channel fault diagnosis information;

[0065] C2: Identify the risk source information in the second emergency collection results, and call the data in the hidden danger database for comparison. According to the comparison results, output the risk source change information before and after the fault, and combine the risk source change information, meteorological data at the time of the fault, and fault characteristics at the fault site to screen the source of the fault.

[0066] In this embodiment, remote sensing satellites are used to collect images regularly and in emergency situations, thus achieving comprehensive monitoring of a wide range of line channel areas. Compared with existing drones and manual inspections, satellite images have the advantages of wide coverage, strong long-term nature, and no environmental or approval restrictions. They are particularly suitable for regular, large-scale inspections and emergency collection before and after disasters. In the specific implementation process, by using the real-time data provided by satellites and meteorology, combined with intelligent analysis technology, the source of foreign matter in the line channel before the disaster can be quickly identified and accurately traced after the failure, making the prediction of potential risks of the line more accurate and comprehensive, and the operation and maintenance personnel can quickly take corresponding measures in advance, greatly reducing the line failure rate caused by uncontrollable factors such as natural disasters, while improving the accuracy of fault analysis and reducing the time for fault handling, thereby ensuring the stable operation of the power system.

[0067] During the implementation process, pre-disaster warning is combined with meteorological warning data and line channel fault diagnosis information to initiate satellite emergency patrols in advance, timely obtain a list of risk sources before potential disasters, and guide on-site disposal; while fault tracing is through emergency patrols of the fault location, combined with historical image data and meteorological information, and multi-period image comparison, which can accurately screen out the source of the fault and help with rapid response and repair. In the fault diagnosis process, in addition to image comparison, the meteorological data at the time of the fault is also integrated to make the fault analysis more comprehensive and accurate. This method makes up for the shortcomings of existing model-based or signal processing technologies that cannot fully cope with complex actual situations.

[0068] The risk source information in this embodiment includes but is not limited to the line name, tower number, coordinates of the center point of the risk source, the distance from the risk source to the nearest line, the direction of the risk source in the line tower, and the risk level of the risk source; as a specific implementation method, the risk level of the risk source is divided into critical risk, general risk, and potential risk. The risk level can be specifically divided according to the disaster information in the meteorological warning information and the distance from the line. Of course, this is only a specific way of dividing the risk level, and other division methods are also within the protection scope of the present invention.

[0069] As a preferred embodiment of the above, Figure 2 As shown, identify the risk source information in each periodic collection result and include the risk source identification results into the hidden danger database, including:

[0070] A21: Identify the risk source information in the satellite images collected in the first phase and include it in the first hidden danger database as the basic data for subsequent risk assessment and comparison;

[0071] A22: Based on the foundation, multi-period image comparison is carried out on the satellite images collected later to obtain information on changes in risk sources and include them in the second hidden danger database;

[0072] A23: The first hidden danger database and the second hidden danger database are merged by using an information matching method to obtain a hidden danger database.

[0073] In this preferred solution, based on the basic hidden danger database of the first phase of satellite images, multi-phase image comparison is carried out to timely discover new, disappeared or changed risk sources, ensuring the dynamic and real-time nature of risk management; through information matching, the first hidden danger database is merged with the hidden danger database collected subsequently to form a comprehensive hidden danger database, recording historical and current risk source information, which is helpful for long-term trend analysis and accurate risk assessment. The entire process is highly automated, from image acquisition and recognition to information comparison and database management, all relying on artificial intelligence and image processing technology to reduce manual intervention and improve efficiency and accuracy.

[0074] As a preferred embodiment of the above, Figure 3 As shown in the figure, based on the foundation, multi-period image comparison is performed on the satellite images collected later to obtain information on changes in risk sources, including:

[0075] A221: Obtain satellite images of two adjacent time phases to be compared as the basis for risk source change analysis. Adjacent time phases refer to satellite images at two time points. The selected time phase interval should be suitable for target detection needs. It should not be too long or too short. Too long intervals may lead to failure to detect changes in risk sources in time, while too short intervals may make it difficult to capture obvious changes. Before comparison, pre-process the images, such as geometric correction and denoising, to ensure the consistency of image data in spatial coordinates and image quality. Perform the following operations on satellite images of each time phase:

[0076] A222: Use the target detection model to obtain the extraction results of each target, and use the semantic segmentation model to obtain the extraction results of the target by surface, and perform vector processing on the two extraction results of each phase; in this step, the target detection model and the semantic segmentation model are used on the satellite image of each phase to obtain different levels of information on the channel risk source, and the target detection model is used to extract each target of the channel risk source and obtain its bounding box. The output boundary is relatively accurate, which is particularly suitable for objects with clear boundaries, such as buildings, equipment, etc. In order to alleviate the problem of missed detection that may occur in target detection, especially channel risk sources with blurred boundary textures, in this preferred solution, a comprehensive judgment is made with the semantic segmentation results; the semantic segmentation model is used to extract each risk source in the channel by surface, and a distribution map of the channel risk source is generated. This method can accurately capture the overall shape of the risk source, and is particularly suitable for larger objects with complex boundaries, such as vegetation, terrain changes, etc. The output is a continuous surface distribution, which can reflect the overall distribution of the risk source in the image, especially when the boundary is unclear or the texture is blurred.

[0077] In this preferred solution, the target detection model can accurately identify the boundary of each target, while the semantic segmentation model can give the overall distribution and shape of the risk source. The combination of the two can complement each other's disadvantages, ensuring that each risk source can be captured and its distribution can be accurately depicted;

[0078] A223: Risk source selection is performed based on vector processing results for different phases;

[0079] A224: Based on the risk source selection in two time phases, obtain the information on risk source changes. Based on the comparison results between the time phases, extract the information on new, disappeared or changed risk sources and update the hidden danger database.

[0080] In this preferred solution, by comparing images of adjacent phases, changes in risk sources can be identified in a timely manner, which helps to predict and deal with hidden dangers before a fault occurs, and significantly improves the fault warning capability of the line.

[0081] As a preferred embodiment of the above, Figure 4 As shown, vector processing is performed on the two extracted results of each phase, including:

[0082] S1: Convert the bounding box extracted by the target detection model into a vector polygon; usually a rectangle represents the location of the risk source, and the four corner points of each bounding box serve as the vertices of the polygon to form a closed rectangular polygon;

[0083] S2: Convert the pixel classification results of the semantic segmentation model into vector polygons; these classification results are usually raster data, and the raster-to-vector conversion algorithm can be used to convert connected pixels of the same type into closed polygons to represent the shape and distribution of each channel risk source;

[0084] In this preferred embodiment, the order of steps S1 and S2 can be swapped or performed simultaneously, which is within the scope of protection of the present invention.

[0085] S3: Perform geometric correction on the vector polygons of target detection and semantic segmentation respectively to achieve alignment with the geographic coordinates of the satellite image. Satellite images may have geometric deformation due to shooting angle, terrain, etc. After geometric correction, the two vector polygons are aligned with the real geographic coordinate system to avoid position deviation problems.

[0086] S4: In the overlapping area of ​​the two vector polygons, the bounding box of the target detection is used to define the scope of the risk source, and for the area missed by the target detection or with blurred boundaries, the pixel distribution result of the semantic segmentation is used to supplement and complete the merging;

[0087] S5: Generate a complete vector data set from the merged vector results, including the shape, center point coordinates, location, boundary and spatial attribute information of each risk source.

[0088] In this preferred solution, the adaptability and stability of the model in complex scenarios are ensured through multi-source fusion. Through regularly updated vector data sets, the addition, disappearance or change of risk sources can be dynamically monitored, and potential risks in line channels can be tracked in real time to improve early warning capabilities. The generated complete vector data set provides basic data for the precise management of risk sources, which can be used for further spatial analysis, risk assessment and decision support. The geometric shapes and spatial attributes of all risk sources are recorded in detail to form a structured and scalable database.

[0089] As a preferred embodiment of the above, Figure 5 As shown, the information matching method includes:

[0090] D1: Extract the center point coordinates of the risk sources in the first hidden danger database and the second hidden danger database respectively, set a spatial distance threshold, such as a few meters or tens of meters, depending on the spatial accuracy requirements of the project, and compare the center point coordinates of each risk source in the two phases. If the distance between the center point coordinates of the two risk sources is less than the spatial distance threshold, mark the same risk source, otherwise mark it as a new or disappeared risk source. Specifically, if there is a risk source in the first hidden danger database but not in the second hidden danger database, it is disappeared, otherwise it is added. This step can quickly narrow the scope of matching objects and improve efficiency;

[0091] D2: Extract boundary polygons for risk sources marked as the same risk source, and calculate the boundary similarity of risk sources in two phases through shape similarity algorithm. The shape similarity algorithm can use Jaccard similarity coefficient or Hausdorff distance, and set boundary similarity threshold, for example, 70%~80%. If the boundary similarity is higher than the boundary similarity threshold, it is marked as the same risk source, otherwise it is marked as a changed risk source;

[0092] D3: For newly added or disappeared risk sources and changed risk sources, extract attribute information and conduct an impact assessment on attribute changes, set an impact threshold, and if the impact of the attribute change is less than the impact threshold, mark it as the same risk source, otherwise keep the original marking status; this step further screens newly added, disappeared or changed risk sources to ensure that changes in attribute information are effectively captured. This is a supplement to steps D1 and D2 to ensure that in addition to spatial and geometric similarity matching, changes in attribute information will not be ignored, especially changes in key attributes such as risk level and type.

[0093] During the implementation process, the changed risk sources are further screened among the marked risk sources of the same type to ensure that even if the risk sources are spatially matched with similarity, the risk sources with changed boundaries or shapes can still be identified through geometric shape similarity detection, thereby reducing misjudgments and missed judgments; only the attribute information of newly added, disappeared and changed risk sources is extracted and compared to avoid unnecessary attribute evaluations of all matching risk sources; by paying attention to the impact of changed risk sources, clearer risk priorities can be given to different risk sources, avoiding overreaction to unimportant or slightly changed risk sources and reducing unnecessary interventions.

[0094] As a preferred embodiment of the above embodiment, the risk source information in the first emergency collection result is identified, and the data in the hidden danger database is called for comparison, and a list of risk sources in the disaster area is output according to the comparison result, including:

[0095] B21: Based on the comparison with the data in the hidden danger database, the risk source information in the first emergency collection result is classified, including the first part that matches the data in the hidden danger database and the second part that does not match;

[0096] B22: Output the list of disaster area risk sources based on the first part.

[0097] In this preferred solution, various dimensions of risk source information are matched, including key attributes such as spatial location, risk level, distance and direction. The system can identify risk sources that are consistent with historical records in the hidden danger database and screen out risk sources that have changed. This multi-dimensional matching method ensures that the identification and classification of risk sources are more accurate and comprehensive.

[0098] In the risk source list, it is preferred to sort by risk level. Taking the level sorting in the above embodiment as an example, the critical risk sources are processed first. For risk sources of the same level, they can be further sorted according to their distance from the line and the direction of the risk source. The line operation and maintenance personnel can send the channel risk source list to the on-site disposal personnel through the information management system. The on-site disposal personnel will carry out targeted cross-channel risk source disposal according to the risk level of the channel risk source, and feedback the disposal results to the information management system through the mobile terminal.

[0099] Among them, it is preferred to include the second part into the hidden danger library so that the historical data can be better applied.

[0100] As a preferred embodiment of the above, the same risk source information identification method is used for the regular collection results, the first emergency collection results, and the second emergency collection results. In specific implementation, the above embodiment can also be used to use the target detection model to obtain the extraction results of each target, and use the semantic segmentation model to obtain the extraction results of the target by surface, and perform vector processing on the two extraction results of each collection result; and identify the risk source based on the vector processing results.

[0101] As a preferred embodiment of the above, obtaining the line channel fault diagnosis information includes:

[0102] Determine the moment when the line fails, and start collecting analysis and judgment factor information at that moment. The analysis and judgment factor information includes at least line data, online monitoring data, meteorological data and hidden danger database data; output fault diagnosis information through an intelligent diagnosis model based on the analysis and judgment factor information.

[0103] In the power system, the moment when a line fails can be determined in a variety of ways, such as fault indicators, online monitoring systems or dispatching system records. The accurate determination of the moment of failure is the basis for subsequent fault diagnosis and analysis. In the implementation process, line data may include line operation data such as current, voltage, phase angle, line load, etc., which can reflect the real-time status of the line; online monitoring data can specifically refer to monitoring equipment installed on power lines or equipment, such as temperature sensors, vibration monitors, etc., which provide real-time data for detecting abnormal conditions of the line; meteorological data includes weather conditions at the time of the failure, such as temperature, humidity, wind speed, rainfall, etc. These data can help analyze the impact of the external environment on the line; hidden danger database data is the historical risk source information recorded in the hidden danger database.

[0104] The collected judgment factor information will be used as input to the intelligent diagnosis model for fault analysis and diagnosis. In specific implementation, the intelligent diagnosis model can use artificial intelligence algorithms, such as machine learning, deep learning and expert system, to automatically analyze the cause and type of fault. This preferred solution integrates a variety of information such as line data, online monitoring data, meteorological data and hidden danger database data, providing rich data support for fault diagnosis and improving the accuracy and comprehensiveness of diagnosis; by starting data collection and intelligent diagnosis at the time of fault, fault information can be captured and analyzed in time, significantly shortening the fault handling time and reducing the impact on power grid operation.

[0105] As a preferred embodiment of the above embodiment, the intelligent diagnosis model is a multimodal fusion model, which can be implemented in the following process:

[0106] Different data sources, including line data, online monitoring data, meteorological data, and hidden danger database data, are regarded as different modalities, and feature extraction is performed on the data of each modality separately, such as using LSTM to process time series data, CNN to process spatial data, and random forest to process numerical data; a fusion layer is used to combine the features extracted by each model, which can be fused by attention mechanism, feature splicing or feature weighted averaging; the fused features are input into a comprehensive classifier, such as a fully connected neural network or an integrated learning model, to output the fault type, location, and possible causes.

[0107] As a preferred embodiment of the above, Figure 6 As shown in the figure, combined with the risk source change information, meteorological data at the time of the fault and the fault characteristics at the fault site, the fault source screening is carried out, including:

[0108] F1: Based on the risk source change information, identify the change type of the risk source and determine the impact of each type of change on the line; in this step, first identify the change type of the risk source before and after the fault. The change type may include new risk sources, such as newly built buildings, the expansion of risk sources, such as the growth of trees, or the disappearance of risk sources; after identifying the change type, evaluate the impact of these changes on the power line. For example, the growth of trees close to the line may increase the risk of touching the line, and newly built tall buildings may interfere with the electromagnetic field of the line. By evaluating the impact of each type of change on the line, the risk sources that may have a significant impact on the line can be preliminarily screened out. This step provides basic data for subsequent analysis and helps narrow the scope of the fault source. Specifically, the impact assessment can be quantified in combination with factors such as the distance, size, and shape changes of the risk source.

[0109] F2: Revise each impact degree based on the meteorological data at the time of the fault and obtain the revised result;

[0110] In specific implementation, the meteorological conditions at the time of the fault will inevitably affect the impact of the risk source on the line. For example, strong winds may cause trees to sway and touch the line, and heavy rain may cause insulator flashover or water accumulation in the line to short-circuit. In this step, the impact level of each type of change is revised by combining the meteorological data at the time of the fault with the impact level of step F1. This revision can more truly reflect the actual threat of the risk source to the line under specific meteorological conditions. The results of the revision can enhance the accuracy of the screening, introduce meteorological factors into the evaluation, and make the identification of the source of the fault more in line with the actual situation. The diversity and combination effects of meteorological data can be considered in the revision process. For example, the simultaneous occurrence of strong winds and rainfall may increase the threat of certain risk sources to the line;

[0111] F3: Based on the revision results and risk source change information, analyze the matching degree between fault characteristics and risk sources, and screen the list of fault sources; matching analysis helps to identify which risk sources are more likely to cause faults, thereby further narrowing the scope of possible fault sources. A variety of methods can be used for matching analysis, such as geographic space analysis based on fault location, time series matching based on current fluctuations, etc.

[0112] F4: Prioritize the screened fault source list and determine the fault source based on the ranking result. The ranking basis in this step may specifically include:

[0113] Distance from the line: The closer the risk source is, the greater the threat it may pose to the line and the higher its priority.

[0114] The degree of impact by meteorological conditions: Risk sources that are significantly affected by severe meteorological conditions should have a higher priority.

[0115] Fault characteristic matching: Risk sources that best match fault characteristics, such as the location and time of fault occurrence, abnormal current or voltage fluctuations, etc., should be given high priority.

[0116] The sorting results can help determine the most likely source of the fault, allowing fault handling personnel to focus on high-priority risk sources. Gradual screening and analysis can systematically identify the source of the fault and accurately locate the cause of the fault by combining meteorological data and fault characteristics. Priority sorting makes fault handling more efficient and provides technical support for the timely repair and safe operation of power lines.

[0117] As a preferred embodiment of the above, Figure 7 As shown, the pre-disaster satellite emergency inspection process in the disaster area and the post-fault emergency inspection in the fault area both use a portable ground station, which is used to upload a single command for image acquisition in the power field and receive satellite images;

[0118] The process of regularly collecting satellite images uses a fixed ground station, which is used to centrally upload image collection instructions in several fields, receive image collection data, and transmit the image collection data to the data center.

[0119] In this preferred solution, the characteristics of long patrol cycles and slow imaging of conventional satellites are broken. The patrol cycle of satellites is reduced from days to hours, which increases the patrol rate of satellites and provides favorable conditions for special patrols before power grid disasters and post-fault analysis and judgment assistance. Portable ground stations can be quickly deployed to the required location to upload a single command and receive satellite images, which is suitable for use in emergency situations. When a quick patrol is required before a disaster or after a fault, the portable ground station can be quickly set up and started, upload image acquisition commands, and obtain satellite images in a short time, supporting timely risk assessment and fault diagnosis. Portable ground stations can be flexibly deployed to different areas, while fixed ground stations can exist as long-term infrastructure. The combination of the two provides good scalability for the system.

[0120] In routine line inspections or periodic risk assessments, fixed ground stations can upload image acquisition instructions in batches. Figure 7 As shown in, it can include environment, agriculture, ocean, electricity and meteorology, reducing the frequency of human intervention and efficiently managing multiple image acquisition tasks; the collected data can be directly transmitted to the data center for centralized processing and analysis, facilitating long-term data accumulation and historical trend analysis.

[0121] Through the implementation of the above optimization plan, by distinguishing the tasks of portable ground stations and fixed ground stations, equipment conflicts of multiple acquisition tasks are avoided and the reliability of system operation is improved. Portable ground stations and fixed ground stations receive and transmit image data respectively, which can realize hierarchical processing and transmission of data, making it easier to distinguish the priority of emergency data from regular data.

[0122] The present invention proposes a deep coupling method of "satellite + meteorology + power grid", which realizes a large-scale survey and control of line channel risk sources, improves the efficiency of channel risk source handling before disasters, changes from passive to active, improves the accuracy of fault analysis, reduces fault handling time, and effectively ensures the safe and stable operation of power grid lines.

[0123] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A line fault diagnosis method based on satellite and meteorology, characterized in that: include: Regularly collect satellite images through remote sensing satellites, and incorporate the regular collection results into the image database, wherein the regular collection is carried out in combination with the power grid line channel control requirements; Identify the risk source information in the periodic collection results, and include the risk source identification results in the hidden danger database; Initiate a pre-disaster satellite emergency patrol process in the disaster area, collect satellite images of the disaster area to obtain a first emergency collection result, and incorporate the first emergency collection result into the image database, wherein the initiation is based on meteorological warning information; Identify the risk source information in the first emergency collection result, call the data in the hidden danger database for comparison, and output a list of risk sources in the disaster area according to the comparison result, wherein the list of risk sources in the disaster area is used to guide the disposal of foreign objects at the channel site; Initiate a post-fault emergency inspection process in the fault area, collect satellite images of the fault area to obtain a second emergency collection result, incorporate the second emergency collection result into the image database, and the initiation is based on line channel fault diagnosis information; Identify the risk source information in the second emergency collection result, call the data in the hidden danger database for comparison, output the risk source change information before and after the fault according to the comparison result, and screen the fault source in combination with the risk source change information, meteorological data at the time of the fault, and fault characteristics at the fault site; Identify the risk source information in the periodic collection results and include the risk source identification results in the hidden danger database, including: Identify the risk source information in the satellite images collected in the first phase and include it in the first hidden danger database as a basis; Based on the above foundation, a multi-period image comparison is performed on the satellite images collected later to obtain information on changes in risk sources and include the information in the second hidden danger database; The first hidden danger database and the second hidden danger database are combined by using an information matching method to obtain the hidden danger database; The information matching method comprises: Extract the center point coordinates of the risk sources in the first hidden danger database and the second hidden danger database respectively, set a spatial distance threshold, and compare the center point coordinates of each risk source in the two phases. If the distance between the center point coordinates of the two risk sources is less than the spatial distance threshold, mark the same risk source, otherwise mark it as a new or disappeared risk source; Extract boundary polygons for the risk sources marked as the same risk source, calculate the boundary similarity of the risk sources in two phases by shape similarity algorithm, set boundary similarity threshold, if the boundary similarity is higher than the boundary similarity threshold, mark them as the same risk source, otherwise mark them as changed risk sources; For the newly added or disappeared risk sources and the changed risk sources, attribute information is extracted and the impact of attribute changes is evaluated, and an impact degree threshold is set. If the impact degree of the attribute change is less than the impact degree threshold, it is marked as the same risk source, otherwise the original marking status is maintained.

2. The satellite and meteorological based line fault diagnosis method according to claim 1, characterized in that: Based on the above foundation, multiple image comparisons are performed on the satellite images collected later to obtain information on changes in risk sources, including: The satellite images of two adjacent time phases to be compared are obtained, and the following operations are performed on the satellite images of each time phase: Use the target detection model to obtain the extraction results of each target, and use the semantic segmentation model to obtain the extraction results of each target by face, and perform vector processing on the two extraction results of each phase; For different phases, risk source selection is performed based on the vector processing result; The risk source selection based on two phases obtains the information of risk source changes.

3. The satellite and meteorological line fault diagnosis method according to claim 2, characterized in that: Vector processing is performed on the two extracted results of each phase, including: Converting the bounding box extracted by the target detection model into a vector polygon; Converting pixel classification results of the semantic segmentation model into vector polygons; Perform geometric correction on the vector polygons of target detection and semantic segmentation respectively to achieve alignment with the geographic coordinates of satellite images; In the overlapping area of ​​the two vector polygons, the scope of the risk source is defined using the bounding box of the target detection, and the pixel distribution result of the semantic segmentation is used to supplement the area missed by the target detection or with blurred boundaries to complete the merging; The merged vector results are used to generate a complete vector dataset, which contains the shape, center point coordinates, location, boundary and spatial attribute information of each risk source.

4. The satellite and meteorological based line fault diagnosis method according to claim 1, characterized in that: Identify the risk source information in the first emergency collection result, call the data in the hidden danger database for comparison, and output the disaster area risk source list according to the comparison result, including: Based on the comparison with the data in the hidden danger database, the risk source information in the first emergency collection result is classified into a first part that matches the data in the hidden danger database and a second part that does not match; Output the list of disaster area risk sources based on the first part.

5. The satellite and meteorological based line fault diagnosis method according to claim 4, characterized in that: The second part is included in the hidden danger library.

6. The satellite and meteorological based line fault diagnosis method according to claim 1, characterized in that: The same risk source information identification method is used for the regular collection results, the first emergency collection results and the second emergency collection results.

7. The satellite and meteorological based line fault diagnosis method according to claim 1, characterized in that: The acquisition of line channel fault diagnosis information includes: Determine the time when the line fails, and start collecting analysis and judgment factor information at the time, wherein the analysis and judgment factor information includes at least line data, online monitoring data, meteorological data and the hidden danger database data; output fault diagnosis information through an intelligent diagnosis model based on the analysis and judgment factor information.

8. The satellite and meteorological based line fault diagnosis method according to claim 7, characterized in that: The intelligent diagnosis model is a multimodal fusion model.

9. The satellite and meteorological based line fault diagnosis method according to claim 1, characterized in that: Combined with the risk source change information, the meteorological data at the time of the fault and the fault characteristics at the fault site, the fault source screening is performed, including: Based on the risk source change information, identify the change type of the risk source, and determine the impact of each type of change on the line; Revising the impact levels based on the meteorological data at the time of the fault to obtain revised results; Based on the revision result and the risk source change information, analyzing the matching degree between the fault characteristics and the risk source, and screening the fault source list; The screened list of fault sources is prioritized, and the fault source is determined based on the ranking result.

10. The satellite and meteorological based line fault diagnosis method according to claim 1, characterized in that: The pre-disaster satellite emergency inspection process of the disaster area and the post-fault emergency inspection of the fault area both use a portable ground station, which is used to upload a single instruction for image acquisition in the power field and receive the satellite image; The process of regularly collecting satellite images adopts a fixed ground station, which is used to centrally upload image collection instructions in several fields, receive image collection data, and transmit the image collection data to a data center.

Images