CN114792319B - Transformer substation inspection method and system based on transformer substation image - Google Patents
Transformer substation inspection method and system based on transformer substation image Download PDFInfo
- Publication number
- CN114792319B CN114792319B CN202210715144.XA CN202210715144A CN114792319B CN 114792319 B CN114792319 B CN 114792319B CN 202210715144 A CN202210715144 A CN 202210715144A CN 114792319 B CN114792319 B CN 114792319B
- Authority
- CN
- China
- Prior art keywords
- inspection
- image
- equipment
- model
- analysis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000007689 inspection Methods 0.000 title claims abstract description 372
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000004458 analytical method Methods 0.000 claims abstract description 121
- 230000009466 transformation Effects 0.000 claims abstract description 76
- 238000013439 planning Methods 0.000 claims abstract description 35
- 238000001514 detection method Methods 0.000 claims abstract description 27
- 230000008569 process Effects 0.000 claims abstract description 14
- 230000007547 defect Effects 0.000 claims description 57
- 230000003631 expected effect Effects 0.000 claims description 38
- 238000007726 management method Methods 0.000 claims description 28
- 238000003860 storage Methods 0.000 claims description 20
- 238000001931 thermography Methods 0.000 claims description 20
- 238000009529 body temperature measurement Methods 0.000 claims description 18
- 230000000694 effects Effects 0.000 claims description 16
- 238000009826 distribution Methods 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 14
- 238000011176 pooling Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 12
- 238000010586 diagram Methods 0.000 claims description 10
- 238000012423 maintenance Methods 0.000 claims description 8
- 238000012552 review Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000033001 locomotion Effects 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 claims description 6
- 230000009471 action Effects 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 238000002372 labelling Methods 0.000 claims description 6
- 239000000741 silica gel Substances 0.000 claims description 6
- 229910002027 silica gel Inorganic materials 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 5
- 238000003745 diagnosis Methods 0.000 claims description 5
- 238000013021 overheating Methods 0.000 claims description 5
- 230000009467 reduction Effects 0.000 claims description 5
- 230000003213 activating effect Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 238000003825 pressing Methods 0.000 claims description 4
- 235000005770 birds nest Nutrition 0.000 claims description 3
- 239000003990 capacitor Substances 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 230000007797 corrosion Effects 0.000 claims description 3
- 238000005260 corrosion Methods 0.000 claims description 3
- 238000002845 discoloration Methods 0.000 claims description 3
- 238000009422 external insulation Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 239000002184 metal Substances 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000001502 supplementing effect Effects 0.000 claims description 3
- 235000005765 wild carrot Nutrition 0.000 claims description 3
- 238000013523 data management Methods 0.000 claims description 2
- 238000013136 deep learning model Methods 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000010438 heat treatment Methods 0.000 description 10
- 230000002159 abnormal effect Effects 0.000 description 8
- 238000004590 computer program Methods 0.000 description 8
- 238000012795 verification Methods 0.000 description 3
- 238000012790 confirmation Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000000779 smoke Substances 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005587 bubbling Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- General Engineering & Computer Science (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Primary Health Care (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a substation inspection method and a substation inspection system based on substation images, and belongs to the technical field of substations. The prior art has low polling efficiency and low accuracy of intelligent model analysis. The invention relates to a substation inspection method based on a substation image, which comprises the steps of constructing a substation inspection target detection model, identifying image information of substation equipment, obtaining types of the substation equipment and component parts thereof in the image information, and obtaining pixel coordinates of the substation equipment and the component parts thereof in an image; and path planning is carried out on the plurality of routing inspection targets through the pixel coordinates to obtain path planning information, the path planning of the power transformation equipment in the shooting process is achieved, and the routing inspection efficiency is effectively improved. Meanwhile, a step adjustment strategy model is constructed, and the inspection image is judged and checked; and according to the judgment and check results, the inspection terminal is adjusted to obtain an expected image meeting the requirements, the picture quality of the power transformation equipment is ensured, and the practical problems of inaccurate image focusing, preset position drifting and the like are effectively avoided.
Description
Technical Field
The invention relates to a substation inspection method and a substation inspection system based on substation images, and belongs to the technical field of substations.
Background
The transformer equipment is an important material basis for power grid safety, along with the continuous expansion of the power grid scale, the contradiction between the equipment scale and the operation and maintenance bearing capacity is increasingly prominent, and a new technology and an operation and maintenance mode need to be explored urgently to improve the operation and maintenance quality and effect.
At present, intelligent inspection device and system such as high definition video camera, robot, unmanned aerial vehicle are by the wide application in transformer substation for replace the manual work to carry out daily work of patrolling and examining, but because its intelligent level is not enough, lead to still need to assist a large amount of manual works to handle in links such as earlier stage configuration, operation maintenance and data analysis of device and system, concrete weak point shows at following problem:
(1) the manual workload required for the configuration and maintenance of the inspection system is huge: in the prior art, a professional is required to comb inspection key points of each device, inspection tasks are configured one by one in a video presetting bit mode, so that the workload of early configuration is huge, for example, a typical 220 KV substation is taken as an example, at least 50 presetting bits are required to be configured to realize intelligent inspection of one transformer, and 1000-2000 presetting bits are required to be configured to the whole substation, which takes about 25 days.
(2) The inspection data quality for intelligent model analysis is low: in the prior art, the main purposes are to capture images and perform static analysis, and typical data sources are cameras, inspection robots and the like. Due to the fact that the holder shakes and the external environment is unpredictable, the quality of a snapshot picture is difficult to guarantee, the actual problems that the focus is inaccurate, the preset position drifts to cause invalidation of the snapshot picture and the like exist, and the accuracy and effectiveness of intelligent model analysis are directly affected.
Further, the invention (publication number: CN 111958594A) discloses a semantic intelligent substation inspection operation robot system and a semantic intelligent substation inspection operation robot method, wherein the semantic intelligent substation inspection operation robot system comprises the following steps: the robot comprises a robot body, wherein a multi-degree-of-freedom mechanical arm is arranged on the robot body, and the tail end of the multi-degree-of-freedom mechanical arm is provided with inspection equipment; the robot body is configured to be capable of autonomously constructing a transformer substation three-dimensional semantic map according to a transformer substation environment and autonomously planning a walking path according to a routing inspection task; the robot body can perform patrol operation according to the planned path; in the process of inspection operation, the pose of the mechanical arm is adjusted according to the position relation between the robot and the equipment to be inspected, so that the image acquisition equipment at the tail end of the mechanical arm can acquire the image of the target inspection equipment at the optimal shooting angle.
According to the scheme, inspection is achieved under the condition that the robot does not stop, the operation mode of 'stopping point-preset' of the traditional transformer substation inspection robot is broken, and the problems that manual configuration of inspection points is large in workload, insufficient in front-end intelligent analysis capacity and low in stopping operation efficiency are solved.
However, the above patent does not disclose how to perform path planning, how to ensure quality of a captured picture, how to avoid practical problems such as inaccurate image focusing and preset bit drift, and the like, and thus, the inspection efficiency and the accuracy and effectiveness of intelligent model analysis are affected.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a method for identifying the image information of the power transformation equipment by constructing a power transformation routing inspection target detection model, obtaining the types of the power transformation equipment and the components thereof in the image information, and acquiring the pixel coordinates of the power transformation equipment and the components thereof in the image; path planning is carried out on the plurality of routing inspection targets through the pixel coordinates to obtain path planning information, path planning in the shooting process of the power transformation equipment is achieved, and routing inspection efficiency is effectively improved; meanwhile, a step adjustment strategy model is constructed, and the inspection image is judged and checked; according to the judgment and check results, the inspection terminal is adjusted to obtain an expected image meeting the requirements, the picture quality of the power transformation equipment is ensured, and the practical problems of inaccurate image focusing, preset position drifting and the like are effectively avoided; and then constructing an associated identification model to identify the expected image and judge the abnormity, outputting and recording the identification result, completing the inspection of the transformer substation, and effectively improving the accuracy and effectiveness of intelligent model analysis.
Aiming at the defects of the prior art, the invention also aims to provide the multispectral inspection terminal which can configure relevant equipment according to a specific inspection target, an inspection angle and an inspection task so as to match with the transformer inspection task, realize the shooting planning of the power transformation equipment and effectively improve the inspection efficiency; meanwhile, the polymorphic analysis module cooperates with the multispectral inspection terminal according to preset point positions to carry out intelligent inspection and analyze multispectral inspection images off-line or on-line to ensure the picture quality of the power transformation equipment; and then carry out the unusual early warning to patrolling and examining the image through station control comprehensive early warning platform, accomplish patrolling and examining of transformer substation, effectively improve transformer substation's patrolling and examining system based on transformer image of the rate of accuracy and validity of intelligent model analysis.
In order to achieve one of the above objects, a first technical solution of the present invention is:
a transformer substation inspection method based on a transformer substation image comprises the following steps:
firstly, controlling an inspection terminal to align a power transformation device at a preset point position to acquire image information;
secondly, constructing a power transformation routing inspection target detection model, identifying the image information in the first step to obtain the types of the power transformation equipment and the components thereof in the image information, and acquiring pixel coordinates of the power transformation equipment and the components thereof in the image;
thirdly, according to the pixel coordinates in the second step, path planning is carried out on the plurality of routing inspection targets to obtain path planning information;
fourthly, sequentially performing inspection on a plurality of inspection targets according to the planning sequence information in the third step to obtain inspection images;
step five, constructing a step adjustment strategy model, and judging and checking the inspection image in the step four; adjusting the inspection terminal according to the judgment and check results to obtain an expected image meeting the requirements;
and sixthly, constructing a relevant identification model to identify the expected image in the fifth step and judge the abnormity, and outputting and recording an identification result to finish the inspection of the transformer substation.
Through continuous exploration and test, a power transformation routing inspection target detection model is constructed, image information of power transformation equipment is identified, the types of the power transformation equipment and the components of the power transformation equipment in the image information are obtained, and pixel coordinates of the power transformation equipment and the components of the power transformation equipment in the image are obtained; and path planning is carried out on the plurality of routing inspection targets through the pixel coordinates to obtain path planning information, path planning in the shooting process of the power transformation equipment is achieved, and routing inspection efficiency is effectively improved.
Meanwhile, a step adjustment strategy model is constructed, and the inspection image is judged and checked; according to the judgment and check results, the inspection terminal is adjusted to obtain an expected image meeting the requirements, the picture quality of the power transformation equipment is ensured, and the practical problems of inaccurate image focusing, preset position drifting and the like are effectively avoided; and then a related identification model is constructed to identify the expected image and judge the abnormity, and the identification result is output and recorded, so that the inspection of the transformer substation is completed, and the accuracy and the effectiveness of intelligent model analysis are effectively improved.
As a preferable technical measure:
in the first step, the method for acquiring the image information comprises the following steps:
in the process of controlling the inspection terminal to align the power transformation equipment, Gaussian modeling is carried out on the moving background of the lens to obtain a Gaussian model which is used for judging whether the lens is in a preset position or not;
the method for judging the Gaussian model comprises the following steps:
s11, preprocessing the input image information, discarding detail features, and keeping macro features to obtain a preprocessed image;
s12, extracting background pixel points from the preprocessed image of S11;
s13, judging whether the variance distribution of the background pixel points in the S12 in a time period accords with Gaussian distribution;
if the Gaussian distribution is met, the lens movement is stopped,
if the Gaussian distribution is not met, the lens is in motion.
As a preferable technical measure:
in the second step, the construction method of the power transformation routing inspection target detection model is as follows:
step 21, collecting and labeling multispectral image data of the power transformation equipment and the components thereof;
the power transformation equipment comprises a transformer or/and a combined electrical appliance or/and a circuit breaker or/and a disconnecting switch or/and a capacitor or/and a reactor or/and a lightning arrester or/and a voltage transformer or/and a current transformer;
the group of components comprise an oil conservator or/and a sleeve or/and a joint or/and a grading ring or/and an oil temperature gauge or/and a pressure gauge;
the multispectral image data comprise a visible light image and an infrared image;
step 22, constructing an attention model to process the multispectral image data in the step 21 so as to eliminate the influence of the complex background of the power target, and enabling the attention of the power transformation inspection target detection model to be concentrated on the power target to obtain an attention image;
step 23, constructing a feature map, processing the attention image in the step 22, and generating a small target prediction image, so that the power transformation routing inspection target detection model is sensitive to a small power target in the attention image;
and step 24, constructing a deep learning model to identify the small target prediction image in the step 23 to obtain identification information.
As a preferable technical measure:
the attention model comprises two independent subunits, which are a channel attention unit and a spatial attention unit respectively;
the channel attention unit is used for extracting the channel attention feature, and the extraction method comprises the following steps:
step 221, obtaining two 1 × 1 × C channel feature maps through maximum pooling and average pooling;
step 222, then sending the channel characteristic diagram in the step 221 into a feedforward artificial neural network model MLP;
step 223, utilizing the output characteristics of the feedforward artificial neural network model MLP in the step 222, and performing pixel-level addition processing to obtain output data;
step 224, activating the output data by using an activation function sigmoid to obtain a final channel attention feature;
the spatial attention unit is used for extracting spatial attention features, and the extraction method comprises the following steps:
s221, taking the channel attention characteristics as input, and performing maximum pooling and average pooling along the channel dimension to obtain two H multiplied by W multiplied by 1 space characteristic graphs;
s223, performing channel cascade on the spatial feature map in the S221, and reducing the dimension to a single channel through a 7 × 7 convolutional layer to obtain dimension reduction data;
s224, activating the dimensionality reduction data by using an activation function sigmoid to obtain a spatial attention feature.
As a preferable technical measure:
in the third step, the path planning method is as follows;
classifying the inspection target according to the category according to the pixel coordinate;
and distributing identification code IDs to a plurality of routing inspection targets of the same type from left to right and from top to bottom.
As a preferable technical measure:
in the fourth step, the expected image is acquired by the following method:
controlling a lens to focus the inspection target according to the position of the inspection target in the image by using a step adjustment strategy model for the xth inspection target in the path planning information, and judging whether the quality of the inspection image achieves the expected effect or not after the lens is adjusted;
if the quality of the inspection image reaches the expected effect, outputting the inspection image as an expected image;
and if the quality of the inspection image does not reach the expected effect, readjusting the lens until the expected effect is reached.
As a preferable technical measure:
the construction method of the step adjustment strategy model comprises the following steps:
step 41, calculating the adjustment step length, calculating according to the step length to obtain a stage expected effect, and generating a corresponding series of lens control actions;
the stage expected effect is the expected values of pixel coordinates, picture proportion and definition of the inspection target after executing each step adjustment strategy;
step 42, adjusting the lens step by step according to the series of lens control actions in the step 41, realizing the routing inspection targets of the front frame and the rear frame through a Hungarian model, performing identity association and registration, and predicting the next appearance coordinate of the routing inspection target through a Kalman model; updating a Kalman prediction function according to the next appearance coordinate;
step 43, after the adjustment in step 42 is completed, comparing the pixel coordinates, the picture ratio and the error between the actual value and the expected value of the definition of the inspection target by adopting a quality judgment model through gradient judgment, and judging whether the actual adjustment effect reaches the expected effect of the stage;
step 44, when the actual adjustment effect of the step 43 does not reach the stage expected effect, executing the step 41-the step 43 to continue the adjustment, and calculating a predicted state value aiming at each inspection target by adopting a foolproof strategy, wherein the state value comprises the size, the position and the screen occupation ratio of a target frame;
performing fool-proofing treatment when the inspection target state cannot be successfully achieved to the expected value after multiple adjustments;
the fool-proof treatment comprises the following contents:
after a plurality of times of trial, giving up the follow-up adjustment of the inspection target, and taking the current state value as the optimal output;
when the actual adjustment effect of the step 43 reaches the stage expected effect, calculating the total actual effect according to the current pixel coordinate, the picture proportion and the definition of the inspection target;
step 45, comparing the overall actual effect in the step 44 with the overall expected effect;
when the overall actual effect reaches the overall expected effect, calling the associated recognition model to perform recognition analysis on the routing inspection target;
when the overall actual effect does not reach the overall expected effect, executing the step 41-the step 43, and carrying out the next long-stage adjustment;
the overall expected effect is the expected values of the pixel coordinates, the picture proportion and the definition of the inspection target after final adjustment is finished.
As a preferable technical measure:
in the fifth step, the associated recognition model comprises a visible light image recognition and analysis unit and an infrared thermal imaging image recognition and analysis unit;
the visible light image recognition and analysis unit comprises the following components:
the method comprises the steps of respectively labeling the equipment types, key parts and typical defects in an image sample library by acquiring and establishing a visible light image sample library of the power transformation equipment and the components thereof, fusing class balanced sampling and center guidance NMS (network management system) strategies based on a labeled sample data set and a corresponding label data set, and training by adopting a neural network with key point identification, semantic segmentation and target detection;
the visible light image recognition and analysis unit comprises a state recognition unit, a meter reading unit and a defect recognition unit:
the state identification unit is used for identifying the states of the power transformation equipment and the components of the power transformation equipment;
the states of the transformation equipment and the components thereof at least comprise a switch blade opening and closing state or/and a breaker opening and closing state or/and a pressing plate opening and closing state;
the meter reading unit is used for identifying meter reading for representing key parameters of the equipment and performing abnormity judgment according to a preset alarm threshold value;
the meter for representing key parameters of the equipment comprises an oil temperature meter or/and a pressure meter SF 6 Or/and a leakage current meter of the lightning arrester;
the defect identification type unit is used for identifying typical defects of the power transformation equipment;
typical defects of the power transformation equipment comprise bird nests or/and foreign matters or/and fireworks or/and external insulation damage or/and silica gel discoloration or/and metal corrosion;
the infrared thermal imaging image recognition and analysis unit comprises the following contents:
acquiring and establishing an infrared thermal imaging image sample library of the power transformation equipment and the components thereof, and labeling the equipment type and the key parts in the image sample library respectively to obtain a sample data set and a corresponding label data set; training a significance detection model based on the labeled sample data set and the corresponding label data set to extract the outline of the inspection target;
and then acquiring a temperature measuring area, and calculating the temperature distribution of the inspection target according to the pixel value in the temperature measuring area, thereby completing the diagnosis of the overheating defect of the inspection target according to the relevant standard specification.
In order to achieve one of the above objects, a second technical solution of the present invention is:
a transformer substation inspection system based on a transformer substation image applies the transformer substation inspection method based on the transformer substation image;
the system comprises a multispectral inspection terminal, a polymorphic analysis module and a station control comprehensive early warning platform;
the multispectral inspection terminal is used for receiving inspection tasks issued by the station control comprehensive early warning platform, matching with the polymorphic analysis module according to preset point positions, developing intelligent inspection, collecting and uploading visible light inspection images and infrared thermal imaging multispectral inspection images;
the multispectral inspection terminal is used for configuring relevant equipment according to a specific inspection target, an inspection angle and an inspection task so as to match the transformer inspection task;
the polymorphic analysis module is used for receiving a patrol task issued by the station control comprehensive early warning platform, cooperating with the multispectral patrol terminal according to preset point positions, developing intelligent patrol, analyzing a visible light patrol image and an infrared thermal imaging multispectral patrol image in an off-line or on-line manner, and uploading an intelligent analysis result to the station control comprehensive early warning platform;
the station control comprehensive early warning platform comprises data management, storage, processing and analysis early warning.
The multi-spectrum routing inspection terminal is provided, and related equipment can be configured according to a specific routing inspection target, a routing inspection angle and a routing inspection task so as to match the transformer routing inspection task, so that shooting planning of the power transformation equipment is realized, and the routing inspection efficiency is effectively improved.
Meanwhile, the polymorphic analysis module is used for cooperating with the multispectral inspection terminal to carry out intelligent inspection according to preset point positions, analyzing the multispectral inspection image off-line or on-line and ensuring the picture quality of the power transformation equipment; and then abnormity early warning is carried out on the inspection image through the station control comprehensive early warning platform, inspection of the transformer substation is completed, and the accuracy and effectiveness of intelligent model analysis are effectively improved.
As a preferable technical measure:
the multispectral inspection terminal is a multispectral camera or/and a pan-tilt camera or/and an inspection robot or/and an unmanned aerial vehicle;
the related equipment is a multispectral pan-tilt camera or/and a multispectral inspection robot or/and an unmanned aerial vehicle;
the multispectral pan-tilt camera is used for developing the tasks of infrared temperature measurement and appearance defect identification of key parts and improving the polling frequency of the key parts of the equipment;
the key part is a high-voltage bushing or/and an oil conservator or/and the top of the body;
the multispectral inspection robot is used for developing meter reading tasks of the oil temperature meter and the oil level meter of the body, typical defect identification tasks and infrared temperature measurement tasks of a plurality of parts, and the inspection precision of equipment is improved;
the typical defect identification task is oil stain on the surface of equipment and the ground or/and color change of silica gel;
the plurality of parts are equipment bodies or/and radiators or/and sleeve lifting seats or/and high-pressure sleeves;
the unmanned aerial vehicle is used for carrying out infrared temperature measurement, state identification and defect identification of a high-voltage sleeve joint or/and a wire clamp or/and an oil level gauge and supplementing a routing inspection blind area of an overhead area of the equipment;
the polymorphic analysis module is reasonably configured according to specific inspection scenes and calculation force requirements and carries a single intelligent chip or a plurality of intelligent chips;
the station control comprehensive early warning platform comprises an inspection task management module, an intelligent model management module, an inspection terminal scheduling module, an inspection result storage management module and a comprehensive analysis early warning module;
the inspection task management module is used for configuring information such as inspection point positions, inspection periods and inspection tasks of the transformer substation and transmitting the information to the multispectral inspection terminal and the polymorphic analysis module;
the intelligent model management module is used for installing, uninstalling and updating various intelligent analysis models and is remotely deployed to the polymorphic analysis module;
the inspection terminal scheduling module is used for receiving the analysis result of the polymorphic analysis module, generating an intelligent rechecking task according to the preset defect grade and the threshold value, transmitting the intelligent rechecking task to the multispectral inspection terminal for rechecking the defect, and transmitting the rechecking result to the inspection result storage management module;
the inspection result storage management module is used for receiving inspection results sent by the polymorphic analysis module and the inspection terminal scheduling module, and the inspection results comprise inspection equipment, inspection time, inspection tasks, inspection images, intelligent analysis results and defect review result information; meanwhile, the inspection result correction function is realized, and the operation and maintenance personnel can check and correct the inspection result through the inspection result storage management module;
the comprehensive analysis early warning module is configured with a multi-terminal data and multi-mode data fusion analysis model, carries out a reading trend analysis task and a three-phase equipment or component temperature measurement result comprehensive analysis task of the same meter based on the patrol data in the patrol result storage and management module, and combines a defect rechecking result to form graded early warning information.
Compared with the prior art, the invention has the following beneficial effects:
through continuous exploration and test, a power transformation routing inspection target detection model is constructed, image information of power transformation equipment is identified, the types of the power transformation equipment and the components of the power transformation equipment in the image information are obtained, and pixel coordinates of the power transformation equipment and the components of the power transformation equipment in the image are obtained; and path planning is carried out on the plurality of routing inspection targets through the pixel coordinates to obtain path planning information, path planning in the shooting process of the power transformation equipment is achieved, and routing inspection efficiency is effectively improved.
Meanwhile, a step adjustment strategy model is constructed, and the inspection image is judged and checked; according to the judgment and check results, the inspection terminal is adjusted to obtain an expected image meeting the requirements, the picture quality of the power transformation equipment is ensured, and the practical problems of inaccurate image focusing, preset position drifting and the like are effectively avoided; and then, a related identification model is constructed to identify the expected image and judge the abnormity, and the identification result is output and recorded, so that the inspection of the transformer substation is completed, and the accuracy and the effectiveness of intelligent model analysis are effectively improved.
Furthermore, the multi-spectrum inspection terminal is arranged, and relevant equipment can be configured according to a specific inspection target, an inspection angle and an inspection task so as to match with the transformer inspection task, so that the shooting planning of the power transformation equipment is realized, and the inspection efficiency is effectively improved.
Furthermore, the polymorphic analysis module is cooperated with the multispectral inspection terminal according to the preset point positions to carry out intelligent inspection, analyze the multispectral inspection image off-line or on-line and ensure the picture quality of the power transformation equipment; and then abnormity early warning is carried out on the inspection image through the station control comprehensive early warning platform, inspection of the transformer substation is completed, and the accuracy and effectiveness of intelligent model analysis are effectively improved.
Drawings
FIG. 1 is a flow chart of the present invention for identifying and determining inspection images;
FIG. 2 is a flow chart of defect review and comprehensive warning according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
On the contrary, the invention is intended to cover alternatives, modifications, equivalents and alternatives which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of the present invention, certain specific details are set forth in order to provide a better understanding of the present invention. It will be apparent to one skilled in the art that the present invention may be practiced without these specific details.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the term "or/and" includes any and all combinations of one or more of the associated listed items.
The invention discloses a transformer substation inspection method based on a transformer substation image, which comprises the following steps:
a transformer substation inspection method based on a transformer substation image comprises the following steps:
firstly, controlling an inspection terminal to align a power transformation device at a preset point position to acquire image information;
secondly, constructing a power transformation routing inspection target detection model, identifying the image information in the first step to obtain the types of the power transformation equipment and the components thereof in the image information, and acquiring pixel coordinates of the power transformation equipment and the components thereof in the image;
thirdly, according to the pixel coordinates in the second step, path planning is carried out on the plurality of routing inspection targets to obtain path planning information;
fourthly, sequentially performing inspection on a plurality of inspection targets according to the planning sequence information in the third step to obtain inspection images;
step five, constructing a step adjustment strategy model, and judging and checking the inspection image in the step four; adjusting the inspection terminal according to the judgment and check results to obtain an expected image meeting the requirements;
and sixthly, constructing a relevant identification model to identify the expected image in the fifth step and judge the abnormity, and outputting and recording an identification result to finish the inspection of the transformer substation.
The invention discloses a transformer substation inspection method based on a transformer substation image, which comprises the following steps:
a transformer substation inspection method based on transformer substation images comprises the following steps:
respectively aiming at visible light and infrared thermal imaging images, configuring a relevant identification model according to the category and preset task type of an inspection target, for example, configuring a sleeve, a grounding lead and other targets in a visual field, configuring an appearance damage and fracture identification model, configuring a passageway, equipment panorama and other models in the visual field, and configuring models of smoke and fire detection, dangerous area invasion and the like;
when the analysis plan arrives, controlling the inspection terminal to align to the inspected equipment according to the preset point position to acquire image information, quickly identifying the type of the power transformation equipment and the component thereof in the image through the power transformation inspection target detection model, and acquiring the pixel coordinates of the power transformation equipment and the component thereof in the image;
automatically planning inspection targets and sequences according to positions of the inspected equipment and key group components thereof in the image, and numbering the inspection targets as 1, 2 and 3 … N in sequence;
step (4) carrying out intelligent inspection on the inspection targets in sequence from the serial number 1 planned in the step (3), controlling the lens to focus on the inspection target according to the position of the inspection target at the x-th position (1 is less than x and less than N) in the planning sequence according to the position of the inspection target in the image, judging whether the image quality reaches the expected effect or not after the lens is adjusted, executing the step (5) if the image quality reaches the expected effect, and readjusting the lens if the image quality does not reach the expected effect until the expected effect is reached;
calling a related identification model to perform intelligent analysis and abnormal judgment on the image of the inspection target, and outputting and recording an analysis result; and if the analysis result needs to be rechecked, linking the preset point positions of other angles, and cooperatively controlling the related patrol terminals to carry out recheck analysis.
After the analysis of the associated recognition model in the step (6) is finished, selecting the (x + 1) th inspection target according to the sequence planned in the step (3), and executing the steps (4) to (5) until the inspection task of the Nth target is executed;
and (7) acquiring image information of a new position according to the preset point position, executing the step (2) to the step (6), and continuously carrying out intelligent routing inspection.
The substation inspection method of the invention combines the substation inspection target and the service characteristics, fuses and utilizes a plurality of AI intelligent models and strategies to form an intelligent inspection technical route of the whole process of inspection image acquisition-quality control-path planning-intelligent analysis, realizes automatic identification equipment and components thereof, automatic routing path planning and automatic adjustment lens to sequentially focus on key inspection points of the equipment, automatically judges and checks the quality of inspection images, automatically dispatches the intelligent model for analysis and abnormal judgment, can set inspection preset positions by using single equipment as a basic inspection unit, reduces the number of the preset positions of the whole substation by more than 5 times, solves the problems of loss of analysis targets, unclear captured images and the like caused by the drift of the preset positions, can meet the automatic positioning and adjustment under complex targets, and improves the quality of the inspection images and the effectiveness of intelligent analysis, improve the quality inspection effect.
One specific embodiment of the associated recognition model of the present invention:
the type of the associated recognition model can be manually configured and associated with a preset point location to construct an analysis point location, and an analysis plan is configured to form an analysis task.
The specific embodiment of the inspection terminal of the invention comprises the following steps:
the inspection terminal of visible light comprises a visible light camera, an unmanned aerial vehicle and a robot,
the thermal imaging inspection terminal is an infrared thermal imaging camera.
The invention discloses a specific embodiment of presetting point positions, which comprises the following steps:
the preset point position is according to the task requirement of patrolling and examining of transformer substation, patrols the point through artifical configuration camera preset position, unmanned aerial vehicle course waypoint, robot.
The invention relates to a specific embodiment for judging the position of a lens, which comprises the following steps:
in the process of controlling the inspection terminal to align the inspected equipment, background Gaussian modeling for lens movement is added to judge whether the lens is in a preset position, so that the lens image movement is prevented from causing interference on the subsequent analysis of the model.
If the lens movement stops, the variance distribution of the pixel points of the background in a time period accords with a Gaussian model, and if the variance distribution of the pixel points of the background in a time period does not accord with the Gaussian model, the variance distribution of the pixel points of the background in the time period does not accord with the Gaussian model.
In order to reduce slight jitter of a lens after a preset position reaches and cause misjudgment of a Gaussian modeling model, resize processing is carried out on an input image, detail features are discarded, and macroscopic features are reserved.
The invention analyzes the specific embodiment of the arrival plan:
the strategies for analyzing the plan arrival are of the following two types:
the first type: the execution plans of different analysis points are not in the same time period, and can be executed in parallel.
The second type: and (3) judging whether resources are sufficient or not when different analysis point locations execute the plan in the same time period, for example, whether the different analysis point locations use the same point location of the same inspection terminal or whether the computational power resource or the decoding capacity of the intelligent analysis module reaches the upper limit, and sequencing through a bubbling model according to a priority strategy under the condition of limited resources to perform time-sharing scheduling.
The invention discloses a specific embodiment of a power transformation inspection target detection model, which comprises the following steps:
the transformer inspection target detection model is pre-established, multi-spectral image data of transformer equipment and components thereof are collected and marked, and training is performed by adopting methods such as deep learning, the inspection target comprises the transformer equipment and the components thereof, the equipment comprises a transformer, a combined electrical appliance, a circuit breaker, an isolating switch, a capacitor, a reactor, a lightning arrester, a voltage transformer, a current transformer and the like, and the components comprise an oil conservator, a sleeve, a joint, an equalizing ring, an oil thermometer, a pressure gauge and the like.
The heating types of the components comprise electric current heating and electric voltage heating.
The current heating group components are as follows: the oil conservator, the box body, a column cap of a 110kV side sleeve (divided into A, B, C phases), a bottom lifting seat of the 110kV side sleeve (divided into A, B, C phases) and a wire clamp.
The components of the electric pressing heater are as follows: 110kV side sleeve (A, B, C phase).
The multispectral image comprises a visible light image, an infrared image and the like. A small target pre-measuring head generated by a low-level high-resolution feature map is added to a power transformation routing inspection target detection model, so that the detection model is more sensitive to tiny power targets in a picture.
An attention module is added on the power transformation routing inspection target detection model to eliminate the influence of the complex background of the power target, so that the attention of the detection model is focused on the interested power target.
The attention module contains two independent sub-modules, a channel attention module and a spatial attention module.
The attention feature extraction method comprises the following steps: two 1 × 1 × C feature maps are obtained through maximum pooling and average pooling, then the feature maps are sent to an MLP, pixel-level addition is carried out on features output by the MLP, and then sigmoid activation is carried out on the features to obtain final channel attention features.
The method for extracting the spatial attention characteristics comprises the following steps: taking one-dimensional channel attention characteristics as input, firstly carrying out maximum pooling and average pooling along the channel dimension to obtain two H multiplied by W multiplied by 1 characteristic graphs, then carrying out channel cascade on the characteristic graphs, then carrying out dimensionality reduction on the characteristic graphs to a single channel through a 7 multiplied by 7 convolutional layer, and finally carrying out sigmoid activation to obtain the spatial attention characteristics.
The invention discloses a specific embodiment for automatically planning inspection:
the method for automatically planning the inspection targets and the sequence comprises the following steps: first, the power targets detected in the screen are classified by category, and then IDs are assigned to a plurality of targets of the same category from left to right and from top to bottom.
The invention discloses a specific embodiment of intelligent inspection:
the intelligent inspection method comprises the following specific steps:
(4-1) adopting a lens refined step adjustment strategy, determining a total expected effect according to the current pixel coordinate, the picture ratio and the definition of the inspection target, completing calculation of adjustment step length through a model, calculating according to the step length to obtain a stage expected effect, and generating a corresponding series of lens control actions, wherein the total expected effect is the expected values of the pixel coordinate, the picture ratio and the definition of the inspection target after final adjustment is completed, and the stage expected effect is the expected values of the pixel coordinate, the picture ratio and the definition of the inspection target after each step adjustment strategy is executed.
(4-2) in the whole lens stepping adjustment process, adopting a target tracking technology, considering the possibility that a scene has a large number of targets and the portability of the AI routing inspection model on a lightweight computing module (the lightweight computing module can only provide a small amount of CPU and GPU computing power), wherein the tracking model selects a traditional model without deep learning, and specifically comprises the following steps: identity association and registration of the polling targets of the front frame and the rear frame are achieved through the Hungarian model, the next appearance coordinate of the polling target is predicted through the Kalman model, and the Kalman prediction function is updated according to the result.
(4-3) aiming at the adjustment of each stage, adopting a quality judgment model, after the staged lens control action is executed, comparing the errors of the actual values of the pixel coordinates, the picture ratio and the definition of the inspection target with the target value through gradient judgment to determine whether to continue fine adjustment at the current stage or enter the adjustment of the next stage, wherein the specific judgment logic and the subsequent execution strategy are divided into three types:
the first type: when the overall expected effect is achieved, calling an intelligent model to analyze the inspection target;
the second type: when the expected effect of the stage is achieved but the total expected effect is not achieved, continuing to enter the next long stage for adjustment;
in the third category: and when the expected effect of the stage is not achieved, adopting a fool-proof strategy, and calculating a predicted state value including the size, the position, the screen occupation ratio and the like of a target frame aiming at each inspection target after the lens is adjusted in the target tracking process. Due to the fact that the camera is disconnected or the ZOOM capacity of the pan-tilt and the lens reaches a boundary, fool-proof processing is conducted when the target state cannot reach a predicted value successfully after multiple adjustments. After a certain number of attempts, the subsequent adjustment of the target is abandoned, and the current result is taken as the best output, so that the condition that the target is in a dead state due to repeated adjustment is avoided, and the adjusted inspection target is the actual maximum expected value.
The invention discloses a specific embodiment of a visible light image identification and analysis unit, which comprises the following steps:
for a visible light image recognition and analysis unit, a visible light image sample library of the power transformation equipment and the components thereof is acquired and established, the equipment type, key parts and typical defects in the image sample library are respectively marked, based on the marked sample data set and a corresponding label data set, category balanced sampling and central instruction NMS (network management system) strategies are fused, neural networks such as key point recognition, semantic segmentation and target detection are adopted for training, and the visible light image recognition and analysis unit is divided into three categories:
the first type: the state identification model is used for identifying the states of equipment and components such as a switch blade opening and closing state, a breaker opening and closing state, a pressing plate opening and closing state and the like;
the second type: the meter reads the class model, realize to oil temperature table, SF 6 Identifying meter readings representing key parameters of equipment such as a pressure meter, a lightning arrester leakage current meter and the like, and performing abnormity judgment according to a preset alarm threshold value;
in the third category: the defect identification model realizes identification of typical defects of the power transformation equipment, such as bird nests, foreign matters, smoke and fire, external insulation damage, silica gel discoloration, metal corrosion and the like.
The invention relates to a specific embodiment of an infrared thermal imaging image recognition analysis unit, which comprises the following steps:
the infrared thermal imaging image identification and analysis unit is used for respectively marking the equipment types and key parts in an image sample library by acquiring and establishing an infrared thermal imaging image sample library of the power transformation equipment and the components thereof, training a significance detection model based on the marked sample data set and a corresponding label data set, realizing the extraction of the outline of the inspection target, further acquiring a temperature measurement area, and calculating the temperature distribution of the inspection target according to the pixel values in the temperature measurement area, thereby completing the diagnosis of the overheating defect of the inspection target according to the relevant standard standards such as DL/T664 electrified equipment infrared diagnosis application standard.
A specific embodiment of the linkage mechanism of the present invention:
the linkage is triggered according to the configuration plan, the input of the plan is keywords described by the analysis result, and the output is the preset point position. After a key description field of an analysis result of a certain analysis point location is captured, a corresponding preset point location is activated according to a plan, and recheck patrol confirmation is carried out from different angles by using different patrol terminals.
The invention discloses a specific embodiment of a substation inspection system, which comprises the following steps:
a transformer substation inspection system based on transformer substation images comprises a multispectral inspection terminal, a polymorphic analysis module and a station control comprehensive early warning platform.
The multispectral inspection terminal is used for receiving an inspection task issued by the station control comprehensive early warning platform, matching with the polymorphic analysis module according to preset point positions, developing AI intelligent inspection, collecting and uploading visible light, infrared thermal imaging and other multispectral inspection images, and the specific forms comprise a multispectral camera, a pan-tilt camera, an inspection robot, an unmanned aerial vehicle and the like.
The multispectral inspection terminal can be reasonably configured according to a specific inspection target, an inspection angle and an inspection task, and the transformer inspection task is taken as an example: a multispectral pan-tilt camera is configured for carrying out infrared temperature measurement and appearance defect identification tasks of key parts such as a high-voltage bushing, an oil conservator and the top of a body, and improving the inspection frequency of the key parts of equipment; the multispectral inspection robot is configured and used for performing meter reading tasks of an oil temperature meter and an oil level meter of the body, typical defect identification tasks of oil stains on the surface of equipment and the ground, color change of silica gel and the like, and infrared temperature measurement tasks of parts of the equipment body, a radiator, a sleeve lifting seat, a high-voltage sleeve and the like, so that the inspection precision of the equipment is improved; an unmanned aerial vehicle is configured for carrying out infrared temperature measurement, state recognition and defect recognition at the positions of a high-voltage sleeve joint, a wire clamp, an oil level gauge and the like, and supplementing inspection blind areas of high-altitude areas of equipment.
The polymorphic analysis module is configured to carry an AI intelligent inspection model and is used for receiving an inspection task issued by the station control comprehensive early warning platform, cooperating with the multispectral inspection terminal according to a preset point position, developing AI intelligent inspection, analyzing the multispectral inspection image off-line or on-line and uploading an intelligent analysis result to the station control comprehensive early warning platform.
The polymorphic analysis module can be reasonably configured according to specific inspection scenes and calculation force requirements. On the one hand, can carry on single intelligent chip, dispose lightweight AI intelligent model, embedding intelligence patrols and examines robot or unmanned aerial vehicle front end and carry out the edge calculation task, promotes the quality that robot or unmanned aerial vehicle gathered the image, improves analysis efficiency. On the other hand, a plurality of intelligent chips can be carried, an AI intelligent model cluster is configured, and the AI intelligent model cluster is arranged at a substation end to develop a centralized analysis task of multiple channels of videos, so that the resource utilization efficiency is improved.
The station control comprehensive early warning platform is composed of an inspection task management module, an intelligent model management module, an inspection terminal scheduling module, an inspection result storage management module and a comprehensive analysis early warning module.
The inspection task management module is used for configuring information such as inspection point positions, inspection periods and inspection tasks of the transformer substation and transmitting the information to the multispectral inspection terminal and the polymorphic analysis module.
And the intelligent model management module is used for installing, uninstalling and updating various AI intelligent analysis models and is remotely deployed to the polymorphic analysis module.
The inspection terminal scheduling module is used for receiving the analysis result of the polymorphic analysis module, generating an intelligent rechecking task according to the preset defect grade and the threshold value, transmitting the intelligent rechecking task to multispectral inspection terminals such as robots and unmanned planes for rechecking the defects, and transmitting the rechecking result to the inspection result storage management module.
The preset defect grade comprises general defects, serious defects and critical defects.
The intelligent review task is divided into an emergency review task and a non-emergency review task.
In a certain inspection task, when a serious defect or an emergency defect is found, the emergency rechecking task is generated immediately, and inspection terminals such as inspection robots and unmanned aerial vehicles and the polymorphic analysis module are dispatched to perform emergency rechecking and intelligent analysis.
The non-emergency rechecking task is characterized in that for all general defects found in a certain inspection task, a non-emergency rechecking list is formed uniformly after the inspection task is completed, and inspection terminals such as inspection robots and unmanned aerial vehicles and polymorphic analysis modules are dispatched to perform one-by-one rechecking and intelligent analysis.
And the inspection result storage and management module is used for receiving inspection results sent by the polymorphic analysis module and the inspection terminal scheduling module, and the inspection results comprise information such as inspection equipment, inspection time, inspection tasks, inspection images, intelligent analysis results and defect review results. Meanwhile, the inspection result correction function is achieved, and the operation and maintenance personnel can check and correct the inspection result through the module.
The comprehensive analysis early warning module is configured with a multi-terminal data and multi-mode data fusion analysis model, stores and manages the inspection data in the module based on the inspection result, carries out tasks such as reading trend analysis of the same meter, comprehensive analysis of temperature measurement results of three-phase equipment or components and the like, and combines the defect rechecking result to form graded early warning information.
And analyzing the reading trend of the same meter, and analyzing and judging whether the change of the reading of the meter is abnormal or not according to the alarm threshold value of the preset trend change and the historical reading.
As shown in fig. 1, a specific embodiment of the present invention for identifying and determining the inspection image:
and inputting the inspection image into a transformer substation inspection system, wherein the transformer substation inspection system divides the inspection image into a visible light image and a thermal imaging image according to the type of the inspection image. And identifying and judging the visible light image, wherein the visible light image identification and judgment comprises state identification, appearance defect identification and personnel behavior judgment. And identifying and judging the thermal imaging image, wherein the identification and judgment result of the thermal imaging image comprises a voltage heating type and a current heating type.
As shown in fig. 2, a specific embodiment of defect review and comprehensive early warning of the present invention:
and the substation inspection system analyzes the inspection image to obtain an analysis result.
And when the analysis result is an emergency defect, generating an emergency rechecking task immediately, and scheduling the polling terminals such as the polling robots and the unmanned aerial vehicles and the polymorphic analysis module to carry out emergency rechecking confirmation. Further, for rechecking the abnormal condition, sending alarm abnormal information; and sending alarm attention information to remind people of needing manual verification when the recheck is normal.
And when the analysis result is a common defect, forming a non-emergency rechecking list, and scheduling the polling terminals such as the polling robots and the unmanned aerial vehicles and the polymorphic analysis modules to recheck one by one. Further, for rechecking the abnormal condition, sending alarm abnormal information; and sending alarm attention information to remind people of needing manual verification when the recheck is normal.
According to the transformer substation inspection system, the transformer substation three-dimensional inspection is realized through the multispectral inspection terminals such as the coordinated dispatching camera, the inspection robot and the unmanned aerial vehicle; the polymorphic analysis module carrying the AI model is flexibly applied, and automatic checking and intelligent analysis of the inspection image quality are realized in a mode of combining edge calculation and centralized analysis; by means of multi-source data comprehensive analysis and multi-terminal cooperative scheduling, comprehensive study and judgment, cooperative rechecking and grading early warning of multispectral inspection defects of the transformer substation are achieved, accuracy of intelligent early warning is improved, and intelligent and practical levels of intelligent inspection of the transformer substation are improved.
The invention relates to an application embodiment for judging overheating abnormity, which comprises the following steps:
the invention is applied to comprehensively analyze the temperature measurement result of the three-phase equipment or component, and respectively carries out the abnormal overheating study and judgment of the voltage heating type equipment and the current heating type equipment according to the related standard specification such as DL/T664 electrified equipment infrared diagnosis application specification and the like and aiming at the three-phase equipment or component by combining the environmental temperature information.
For the current heating type equipment, two modes of relative temperature difference and absolute temperature are adopted for parallel judgment, and the calculation formula of the relative temperature difference is as follows:δ= (T1-T2)/(T1-T0) × 100%, where T0= ambient temperature of the equipment area, T1= highest temperature in the temperature measurement frame, and T2= lowest value between T1 of the same temperature measurement points of the same three-phase equipment (for three-phase split equipment) or lowest value of temperature in the same temperature measurement frame (for three-phase non-split equipment, such as a body, a conservator, an isolation switch blade, etc.); and the other is voltage heating type equipment, the inspection conclusion of which is given based on the comparison result of the difference value between the highest temperature and the lowest temperature in the temperature measuring frame and the threshold, and the specific study and judgment logic is shown in table 1.
TABLE 1
The grading early warning information is divided into attention and warning, and warning information is sent when the rechecking judgment is that the defect exists; and judging the defect in the first inspection, judging the defect to be normal after rechecking, sending attention information and reminding the need of manual verification.
An embodiment of an apparatus to which the method of the invention is applied:
a computer apparatus, comprising:
one or more processors;
storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement a substation inspection method based on substation images as described above.
An embodiment of a computer medium to which the method of the invention is applied is:
a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a substation inspection method based on substation images as described above.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as methods, systems, computer program products. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting the same, and although the present invention is described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.
Claims (3)
1. A transformer substation inspection method based on transformer substation images is characterized in that,
the method comprises the following steps:
firstly, controlling an inspection terminal to align to a power transformation device at a preset point position to acquire image information;
secondly, constructing a power transformation routing inspection target detection model, identifying the image information in the first step to obtain the types of the power transformation equipment and the components thereof in the image information, and acquiring pixel coordinates of the power transformation equipment and the components thereof in the image;
thirdly, according to the pixel coordinates in the second step, path planning is carried out on the plurality of routing inspection targets to obtain path planning information;
fourthly, sequentially performing inspection on a plurality of inspection targets according to the planning sequence information in the third step to obtain inspection images;
step five, constructing a step adjustment strategy model, and judging and checking the inspection image in the step four; adjusting the inspection terminal according to the judgment and check results to obtain an expected image meeting the requirements;
sixthly, constructing a relevant identification model to identify the expected image in the fifth step and judge the abnormity, and outputting and recording an identification result to finish the inspection of the transformer substation;
in the first step, the method for acquiring the image information comprises the following steps:
in the process of controlling the inspection terminal to align the power transformation equipment, Gaussian modeling is carried out on the moving background of the lens to obtain a Gaussian model which is used for judging whether the lens is in a preset position or not;
the method for judging the Gaussian model comprises the following steps:
s11, preprocessing the input image information, discarding detail features, and keeping macro features to obtain a preprocessed image;
s12, extracting background pixel points from the preprocessed image of S11;
s13, judging whether the variance distribution of the background pixel points in the S12 in a time period accords with Gaussian distribution;
if the Gaussian distribution is met, the lens movement is stopped,
if the shot does not accord with the Gaussian distribution, the shot is in motion;
in the second step, the construction method of the power transformation patrol inspection target detection model comprises the following steps:
step 21, collecting and labeling multispectral image data of the power transformation equipment and the components thereof;
the power transformation equipment comprises a transformer or/and a combined electrical appliance or/and a circuit breaker or/and a disconnecting switch or/and a capacitor or/and a reactor or/and a lightning arrester or/and a voltage transformer or/and a current transformer;
the group of components comprise an oil conservator or/and a sleeve or/and a joint or/and a grading ring or/and an oil temperature gauge or/and a pressure gauge;
the multispectral image data comprises a visible light image and an infrared image;
step 22, constructing an attention model to process the multispectral image data in the step 21 so as to eliminate the influence of the complex background of the power target, and enabling the attention of the power transformation patrol inspection target detection model to be concentrated on the power target to obtain an attention image;
step 23, constructing a feature map, processing the attention image in the step 22, and generating a small target prediction image, so that the power transformation routing inspection target detection model is sensitive to a small power target in the attention image;
step 24, constructing a deep learning model to identify the small target prediction image in the step 23 to obtain identification information;
the attention model comprises two independent subunits, which are a channel attention unit and a spatial attention unit respectively;
the channel attention unit is used for extracting the channel attention feature, and the extraction method comprises the following steps:
step 221, obtaining two 1 × 1 × C channel feature maps through maximum pooling and average pooling;
step 222, then sending the channel characteristic diagram in the step 221 into a feedforward artificial neural network model MLP;
step 223, utilizing the output characteristics of the feedforward artificial neural network model MLP in the step 222, and performing pixel-level addition processing to obtain output data;
step 224, activating the output data by using an activation function sigmoid to obtain a final channel attention feature;
the spatial attention unit is used for extracting spatial attention features, and the extraction method comprises the following steps:
s221, taking the channel attention characteristics as input, and performing maximum pooling and average pooling along the channel dimension to obtain two H multiplied by W multiplied by 1 space characteristic graphs;
s223, performing channel cascade on the spatial feature map in the S221, and reducing the dimension to a single channel through a 7 × 7 convolutional layer to obtain dimension reduction data;
s224, activating the dimensionality reduction data by using an activation function sigmoid to obtain a space attention feature;
in the third step, the path planning method is as follows;
classifying the inspection target according to the category according to the pixel coordinate;
distributing identification code IDs to a plurality of routing inspection targets of the same type from left to right and from top to bottom;
in the fifth step, the expected image is acquired by the following method:
controlling a lens to focus the inspection target according to the position of the inspection target in the image by using a step adjustment strategy model for the xth inspection target in the path planning information, and judging whether the quality of the inspection image achieves the expected effect or not after the lens is adjusted;
if the quality of the inspection image reaches the expected effect, outputting the inspection image as an expected image;
if the quality of the inspection image does not reach the expected effect, readjusting the lens until the expected effect is reached;
the construction method of the step adjustment strategy model comprises the following steps:
step 41, calculating the adjustment step length, calculating according to the step length to obtain a stage expected effect, and generating a corresponding series of lens control actions;
the stage expected effect is the expected values of pixel coordinates, picture proportion and definition of the inspection target after executing each step adjustment strategy;
step 42, adjusting the lens step by step according to the series of lens control actions in the step 41, realizing the routing inspection targets of the front frame and the rear frame through a Hungarian model, performing identity association and registration, and predicting the next appearance coordinate of the routing inspection target through a Kalman model; updating a Kalman prediction function according to the next appearance coordinate;
step 43, after the adjustment in step 42 is completed, comparing the pixel coordinates, the picture ratio and the error between the actual value and the expected value of the definition of the inspection target by adopting a quality judgment model through gradient judgment, and judging whether the actual adjustment effect reaches the expected effect of the stage;
step 44, when the actual adjustment effect in the step 43 does not reach the stage expected effect, executing the step 41 to the step 43 to continue the adjustment, and adopting a fool-proof strategy to calculate a predicted state value aiming at each inspection target, wherein the state value comprises the size, the position and the screen occupation ratio of a target frame;
performing fool-proofing treatment when the inspection target state cannot be successfully achieved to the expected value after multiple adjustments;
the fool-proof treatment comprises the following contents:
after trying for a plurality of times, giving up the subsequent adjustment of the inspection target, and taking the current state value as the optimal output;
when the actual adjusting effect of the step 43 reaches the stage expected effect, calculating the total actual effect according to the current pixel coordinate, the picture ratio and the definition of the inspection target;
step 45, comparing the overall actual effect in the step 44 with the overall expected effect;
when the overall actual effect reaches the overall expected effect, calling the associated recognition model to perform recognition analysis on the routing inspection target;
when the overall actual effect does not reach the overall expected effect, executing the step 41-the step 43, and carrying out the next long-stage adjustment;
the overall expected effect is an expected value of pixel coordinates, picture proportion and definition of the inspection target after final adjustment is finished;
in the sixth step, the associated recognition model comprises a visible light image recognition and analysis unit and an infrared thermal imaging image recognition and analysis unit;
the visible light image recognition and analysis unit comprises the following components:
the method comprises the steps of respectively labeling the equipment types, key parts and typical defects in an image sample library by acquiring and establishing a visible light image sample library of the power transformation equipment and the components thereof, fusing class balanced sampling and center guidance NMS (network management system) strategies based on a labeled sample data set and a corresponding label data set, and training by adopting a neural network with key point identification, semantic segmentation and target detection;
the visible light image recognition and analysis unit comprises a state recognition unit, a meter reading unit and a defect recognition unit:
the state identification unit is used for identifying the states of the power transformation equipment and the components thereof;
the states of the transformation equipment and the components thereof at least comprise a switch blade opening and closing state or/and a breaker opening and closing state or/and a pressing plate opening and closing state;
the meter reading unit is used for identifying meter reading for representing key parameters of the equipment and performing abnormity judgment according to a preset alarm threshold value;
the meter for representing key parameters of the equipment comprises an oil temperature meter or/and a pressure meter SF 6 Or/and a leakage current meter of the lightning arrester;
the defect identification type unit is used for identifying typical defects of the power transformation equipment;
typical defects of the power transformation equipment comprise bird nests or/and foreign matters or/and fireworks or/and external insulation damage or/and silica gel discoloration or/and metal corrosion;
the infrared thermal imaging image recognition analysis unit comprises the following contents:
acquiring and establishing an infrared thermal imaging image sample library of the power transformation equipment and the components thereof, and labeling the equipment type and the key parts in the image sample library respectively to obtain a sample data set and a corresponding label data set; training a significance detection model based on the labeled sample data set and the corresponding label data set to extract the outline of the inspection target;
and then acquiring a temperature measuring area, and calculating the temperature distribution of the inspection target according to the pixel value in the temperature measuring area, thereby completing the diagnosis of the overheating defect of the inspection target according to the relevant standard specification.
2. A transformer substation inspection system based on transformer substation images is characterized in that,
applying the substation inspection method based on the substation image according to claim 1;
the system comprises a multispectral inspection terminal, a polymorphic analysis module and a station control comprehensive early warning platform;
the multispectral inspection terminal is used for receiving inspection tasks issued by the station control comprehensive early warning platform, matching with the polymorphic analysis module according to preset point positions, developing intelligent inspection, collecting and uploading visible light inspection images and infrared thermal imaging multispectral inspection images;
the multispectral inspection terminal is used for configuring relevant equipment according to a specific inspection target, an inspection angle and an inspection task so as to match the transformer inspection task;
the polymorphic analysis module is used for receiving a patrol task issued by the station control comprehensive early warning platform, cooperating with the multispectral patrol terminal according to preset point positions, developing intelligent patrol, analyzing a visible light patrol image and an infrared thermal imaging multispectral patrol image in an off-line or on-line manner, and uploading an intelligent analysis result to the station control comprehensive early warning platform;
the station control comprehensive early warning platform comprises data management, storage, processing and analysis early warning.
3. The substation inspection system based on substation images according to claim 2,
the multispectral inspection terminal is a multispectral camera or/and a pan-tilt camera or/and an inspection robot or/and an unmanned aerial vehicle;
the related equipment is a multispectral pan-tilt camera or/and a multispectral inspection robot or/and an unmanned aerial vehicle;
the multispectral pan-tilt camera is used for developing the tasks of infrared temperature measurement and appearance defect identification of key parts and improving the polling frequency of the key parts of the equipment;
the key part is a high-voltage bushing or/and an oil conservator or/and the top of the body;
the multispectral inspection robot is used for developing meter reading tasks of the body oil temperature meter and the oil level meter, typical defect identification tasks and infrared temperature measurement tasks of a plurality of parts;
the typical defect identification task is oil stain on the surface of equipment and the ground or/and color change of silica gel;
the plurality of parts are equipment bodies or/and radiators or/and sleeve lifting seats or/and high-pressure sleeves;
the unmanned aerial vehicle is used for carrying out infrared temperature measurement, state identification and defect identification of a high-voltage sleeve joint or/and a wire clamp or/and an oil level gauge and supplementing a routing inspection blind area of an overhead area of the equipment;
the polymorphic analysis module is reasonably configured according to specific inspection scenes and calculation force requirements and carries a single intelligent chip or a plurality of intelligent chips;
the station control comprehensive early warning platform comprises an inspection task management module, an intelligent model management module, an inspection terminal scheduling module, an inspection result storage management module and a comprehensive analysis early warning module;
the inspection task management module is used for configuring information such as inspection point positions, inspection periods and inspection tasks of the transformer substation and transmitting the information to the multispectral inspection terminal and the polymorphic analysis module;
the intelligent model management module is used for installing, uninstalling and updating various intelligent analysis models and is remotely deployed to the polymorphic analysis module;
the inspection terminal scheduling module is used for receiving the analysis result of the polymorphic analysis module, generating an intelligent rechecking task according to the preset defect grade and the threshold value, transmitting the intelligent rechecking task to the multispectral inspection terminal for rechecking the defect, and transmitting the rechecking result to the inspection result storage management module;
the inspection result storage management module is used for receiving inspection results sent by the polymorphic analysis module and the inspection terminal scheduling module, and the inspection results comprise inspection equipment, inspection time, inspection tasks, inspection images, intelligent analysis results and defect review result information; meanwhile, the inspection result correction function is realized, and the operation and maintenance personnel can check and correct the inspection result through the inspection result storage management module;
the comprehensive analysis early warning module is configured with a multi-terminal data and multi-mode data fusion analysis model, carries out a same meter reading trend analysis task and a three-phase equipment or component temperature measurement result comprehensive analysis task based on the inspection data in the inspection result storage management module, and combines a defect rechecking result to form graded early warning information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210715144.XA CN114792319B (en) | 2022-06-23 | 2022-06-23 | Transformer substation inspection method and system based on transformer substation image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210715144.XA CN114792319B (en) | 2022-06-23 | 2022-06-23 | Transformer substation inspection method and system based on transformer substation image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114792319A CN114792319A (en) | 2022-07-26 |
| CN114792319B true CN114792319B (en) | 2022-09-20 |
Family
ID=82463630
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210715144.XA Active CN114792319B (en) | 2022-06-23 | 2022-06-23 | Transformer substation inspection method and system based on transformer substation image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114792319B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115150559B (en) * | 2022-09-06 | 2022-11-25 | 国网天津市电力公司高压分公司 | Remote vision system with acquisition self-adjustment calculation compensation and calculation compensation method |
| CN115202404B (en) * | 2022-09-15 | 2022-12-02 | 广东容祺智能科技有限公司 | Maintenance and inspection method for photovoltaic power generation set based on unmanned aerial vehicle |
| CN115348393B (en) * | 2022-10-20 | 2023-02-03 | 慧视云创(北京)科技有限公司 | Automatic setting method for preset position of camera device and camera group |
| CN115760854A (en) * | 2023-01-09 | 2023-03-07 | 佰聆数据股份有限公司 | Deep learning-based power equipment defect detection method and device and electronic equipment |
| CN116228778B (en) * | 2023-05-10 | 2023-09-08 | 国网山东省电力公司菏泽供电公司 | Insulator rupture detection method and system based on multi-mode information fusion |
| CN116629842B (en) * | 2023-07-19 | 2023-11-07 | 国网浙江省电力有限公司苍南县供电公司 | Power equipment inspection method and platform based on image processing |
| CN117312591B (en) * | 2023-10-17 | 2024-03-12 | 南京海汇装备科技有限公司 | Image data storage management system and method based on virtual reality |
| CN117216673B (en) * | 2023-11-08 | 2024-03-12 | 国网江西省电力有限公司电力科学研究院 | Current transformer monitoring evaluation overhauls platform |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109635720A (en) * | 2018-12-10 | 2019-04-16 | 江南大学 | The illegal road occupying real-time detection method actively monitored based on video |
| CN112102369A (en) * | 2020-09-11 | 2020-12-18 | 陕西欧卡电子智能科技有限公司 | Autonomous inspection method, device and equipment for water surface floating target and storage medium |
| CN112733824A (en) * | 2021-04-06 | 2021-04-30 | 中国电力科学研究院有限公司 | Transformer equipment defect diagnosis method and system based on video image intelligent front end |
| CN112904877A (en) * | 2021-01-14 | 2021-06-04 | 星闪世图(台州)科技有限公司 | Automatic fan blade inspection system and method based on unmanned aerial vehicle |
| CN113139521A (en) * | 2021-05-17 | 2021-07-20 | 中国大唐集团科学技术研究院有限公司中南电力试验研究院 | Pedestrian boundary crossing monitoring method for electric power monitoring |
| CN113590878A (en) * | 2021-07-28 | 2021-11-02 | 宜宾中星技术智能系统有限公司 | Method and device for planning path on video picture and terminal equipment |
| CN114332697A (en) * | 2021-12-19 | 2022-04-12 | 西安科技大学 | Method, system, equipment and medium for detecting faults of multiple types of targets in power transmission line |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103455847B (en) * | 2012-05-30 | 2017-04-12 | 国际商业机器公司 | Method and device for determining paths |
| CN111259809B (en) * | 2020-01-17 | 2021-08-17 | 五邑大学 | Unmanned aerial vehicle coastline floating garbage inspection system based on DANet |
| CN112350441B (en) * | 2020-11-03 | 2022-10-14 | 国网智能科技股份有限公司 | Online intelligent inspection system and method for transformer substation |
-
2022
- 2022-06-23 CN CN202210715144.XA patent/CN114792319B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109635720A (en) * | 2018-12-10 | 2019-04-16 | 江南大学 | The illegal road occupying real-time detection method actively monitored based on video |
| CN112102369A (en) * | 2020-09-11 | 2020-12-18 | 陕西欧卡电子智能科技有限公司 | Autonomous inspection method, device and equipment for water surface floating target and storage medium |
| CN112904877A (en) * | 2021-01-14 | 2021-06-04 | 星闪世图(台州)科技有限公司 | Automatic fan blade inspection system and method based on unmanned aerial vehicle |
| CN112733824A (en) * | 2021-04-06 | 2021-04-30 | 中国电力科学研究院有限公司 | Transformer equipment defect diagnosis method and system based on video image intelligent front end |
| CN113139521A (en) * | 2021-05-17 | 2021-07-20 | 中国大唐集团科学技术研究院有限公司中南电力试验研究院 | Pedestrian boundary crossing monitoring method for electric power monitoring |
| CN113590878A (en) * | 2021-07-28 | 2021-11-02 | 宜宾中星技术智能系统有限公司 | Method and device for planning path on video picture and terminal equipment |
| CN114332697A (en) * | 2021-12-19 | 2022-04-12 | 西安科技大学 | Method, system, equipment and medium for detecting faults of multiple types of targets in power transmission line |
Non-Patent Citations (1)
| Title |
|---|
| 基于多源异构视觉的变电站机器人巡检技术研究;梁松伟;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20220215;摘要、正文第1-75页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114792319A (en) | 2022-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114792319B (en) | Transformer substation inspection method and system based on transformer substation image | |
| CN111897332B (en) | Semantic intelligent substation robot humanoid inspection operation method and system | |
| CN109977813B (en) | Inspection robot target positioning method based on deep learning framework | |
| CN109508580B (en) | Traffic signal lamp identification method and device | |
| CN111958592B (en) | Image semantic analysis system and method for transformer substation inspection robot | |
| CN110266938B (en) | Transformer substation equipment intelligent shooting method and device based on deep learning | |
| CN112990310A (en) | Artificial intelligence system and method for serving electric power robot | |
| CN109858367B (en) | Visual automatic detection method and system for worker through supporting unsafe behaviors | |
| CN110703800A (en) | Unmanned aerial vehicle-based intelligent identification method and system for electric power facilities | |
| CN108734143A (en) | A kind of transmission line of electricity online test method based on binocular vision of crusing robot | |
| CN112668696A (en) | Unmanned aerial vehicle power grid inspection method and system based on embedded deep learning | |
| CN110567964A (en) | method and device for detecting defects of power transformation equipment and storage medium | |
| CN111988524A (en) | Unmanned aerial vehicle and camera collaborative obstacle avoidance method, server and storage medium | |
| CN115649501B (en) | Unmanned aerial vehicle night lighting system and method | |
| CN109542114A (en) | A kind of unmanned plane polling transmission line method and system | |
| CN109389322A (en) | The disconnected broken lot recognition methods of grounded-line based on target detection and long memory models in short-term | |
| CN115169602A (en) | Maintenance method and device for power equipment, storage medium and computer equipment | |
| CN117589177B (en) | Autonomous navigation method based on industrial large model | |
| Manninen et al. | Multi-stage deep learning networks for automated assessment of electricity transmission infrastructure using fly-by images | |
| CN113076808B (en) | Method for accurately acquiring bidirectional traffic flow through image algorithm | |
| CN112542800A (en) | Method and system for identifying transmission line fault | |
| CN114167245B (en) | Intelligent detection method for partial discharge on surface of power transmission and transformation equipment and unmanned aerial vehicle fusion ultraviolet system | |
| CN116055521A (en) | Inspection system and image recognition method for electric inspection robot | |
| CN115457656A (en) | Method, device and equipment for determining operation duration and storage medium | |
| CN114038040A (en) | Machine room inspection monitoring method, device and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |