CN116486240A - Application of image recognition algorithm in intelligent inspection method of unmanned aerial vehicle of power transmission line - Google Patents
Application of image recognition algorithm in intelligent inspection method of unmanned aerial vehicle of power transmission line Download PDFInfo
- Publication number
- CN116486240A CN116486240A CN202310500114.1A CN202310500114A CN116486240A CN 116486240 A CN116486240 A CN 116486240A CN 202310500114 A CN202310500114 A CN 202310500114A CN 116486240 A CN116486240 A CN 116486240A
- Authority
- CN
- China
- Prior art keywords
- analyzing
- analysis
- model
- image
- equipment
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 110
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 41
- 238000007689 inspection Methods 0.000 title claims abstract description 40
- 230000005540 biological transmission Effects 0.000 title claims abstract description 32
- 238000004458 analytical method Methods 0.000 claims abstract description 70
- 238000005516 engineering process Methods 0.000 claims abstract description 54
- 238000010801 machine learning Methods 0.000 claims abstract description 28
- 238000001514 detection method Methods 0.000 claims abstract description 23
- 238000002372 labelling Methods 0.000 claims abstract description 21
- 230000008859 change Effects 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 18
- 230000000007 visual effect Effects 0.000 claims abstract description 18
- 238000007781 pre-processing Methods 0.000 claims abstract description 13
- 238000000605 extraction Methods 0.000 claims abstract description 11
- 230000007246 mechanism Effects 0.000 claims abstract description 5
- 238000013145 classification model Methods 0.000 claims description 9
- 238000013136 deep learning model Methods 0.000 claims description 8
- 238000007726 management method Methods 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 6
- 238000005065 mining Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 5
- 238000013499 data model Methods 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 238000007621 cluster analysis Methods 0.000 claims description 3
- 238000003066 decision tree Methods 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 claims description 3
- 239000002243 precursor Substances 0.000 claims description 3
- 238000011084 recovery Methods 0.000 claims description 3
- 238000007670 refining Methods 0.000 claims description 3
- 238000010187 selection method Methods 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 2
- 238000013178 mathematical model Methods 0.000 claims description 2
- 238000007619 statistical method Methods 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000004090 dissolution Methods 0.000 abstract 1
- 239000012188 paraffin wax Substances 0.000 abstract 1
- 230000033001 locomotion Effects 0.000 description 7
- 230000008901 benefit Effects 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000013523 data management Methods 0.000 description 2
- 238000007418 data mining Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000006424 Flood reaction Methods 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000012880 independent component analysis Methods 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000007639 printing Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/94—Hardware or software architectures specially adapted for image or video understanding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
- G06V10/765—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/17—Terrestrial scenes taken from planes or by drones
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Remote Sensing (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
Abstract
An application of an image recognition algorithm in an intelligent inspection method of an electric transmission line unmanned aerial vehicle relates to the technical field of paraffin dissolution, and comprises the following steps: analyzing a preprocessing method of the video image; and a second step of: multimedia recognition and stream processing machine learning model analysis; and a third step of: analyzing and applying equipment map state change detection technology; the invention has the beneficial effects that: the method comprises the steps that an image feature description algorithm is provided through video images acquired through unmanned aerial vehicle inspection based on sparse coding, visual words, feature appearance extraction and other technologies, and structured description of unstructured image data is achieved; providing an object labeling technology based on a machine learning model under a target visual mechanism, and realizing an object recognition and classification method; a soft change detection technology is provided to realize the detection of the state change of the equipment.
Description
Technical Field
The invention relates to the technical field of intelligent inspection methods of unmanned aerial vehicles of power transmission lines, in particular to application of an image recognition algorithm to the intelligent inspection method of the unmanned aerial vehicles of the power transmission lines;
background
The traditional manual inspection method is large in workload and difficult in conditions, and particularly is used for inspecting transmission lines in mountainous areas and across large rivers, and inspecting lines during ice disasters, floods, earthquakes, landslides and nights, so that the time is long, the labor cost is high, the difficulty is high, and the risk is high; in recent years, the technology for realizing surface observation, geological exploration and line inspection by adopting a low-altitude remote sensing mode of an unmanned helicopter is mature gradually, and particularly, the unmanned helicopter inspection gradually becomes an important development direction of high-voltage transmission line inspection by the characteristics of high efficiency, accuracy, safety and the like, and faults and defects of the transmission line can be found by processing and analyzing a large number of aerial images acquired by aerial inspection, and electric power facility disasters in special geological environments are monitored and early warned; however, the number of images or video data acquired in the aerial inspection process is very large, and if the modes of post manual screening, manual analysis and judgment are adopted, huge video and picture information is directly processed, the work is undoubtedly very difficult, and key information is easily omitted; if the abnormity of the power transmission and transformation line is monitored on line, judged and found in real time by adopting manpower, the power transmission and transformation line is more laborious work and key information is easier to miss and misreport.
Disclosure of Invention
In order to solve the problems, the invention provides an application of an image recognition algorithm to an intelligent inspection method of an unmanned aerial vehicle of an electric transmission line, which is characterized in that: the first step: preprocessing method analysis of video image
The preprocessing method for the inspection image is analyzed, and a preprocessing algorithm with strong universality, good processing effect and wide application range is found, particularly in the aspects of brightness balance, contrast enhancement, image denoising and deblurring; for the anomaly identification of the aerial transmission line inspection image sequence, key frame extraction is also needed, so that the image information operand is reduced, and the identification accuracy is improved;
and a second step of: multimedia recognition and stream processing machine learning model analysis
Analyzing and establishing a typical scene learning database required by supervised, semi-supervised and weakly supervised machine learning models; an analysis database comprising an analysis of database model building methods; collecting original data; the functions of adding and deleting data in the database are realized; analyzing digital feature extraction and understanding of the image data; analyzing a sample marking method;
analyzing and establishing a cascade classification model, a probability map model and a deep learning model machine learning model, and verifying;
in the cascade classification model, analyzing cascade strategies of each base learner; analyzing a learning strategy and an reasoning algorithm of the base learner; a confidence threshold setting method of an analysis base learner; verifying and adjusting strategies of the analysis model;
in the establishment of the probability map model, analyzing the relation among random variables; analyzing the directed graph or the undirected graph, and generating a model or a selection method of a discrimination model; verifying and adjusting strategies of the analysis model;
analyzing the deep learning model, comprising: analyzing the hierarchy and mechanism of learners in deep learning to determine model structure and complexity; a method for analyzing the memory consumption of the system and reducing the complexity of the algorithm; verifying and adjusting strategies of the analysis model;
analyzing and verifying a learning method of machine learning model parameters;
a method for establishing an analysis objective function expression; an equation or inequality abstract expression method for analyzing constraint conditions; the convergence and the convergence speed of a parameter optimization algorithm in actual application are inspected;
analyzing and verifying an inference method based on a machine learning model;
performing analysis of a supervised learning method: a marking method for analyzing training samples required by the supervised learning method; analyzing and selecting reasonable sample quantity; analyzing an convincing inferred result evaluation method; specifically analyzing decision trees, boosting and Bagging algorithms, artificial neural networks and popular and effective methods of support vector machines;
performing an analysis of an unsupervised learning method; a method of analyzing a total data distribution inferred from sample information; a main analysis cluster analysis and association rule analysis algorithm;
performing analysis of a semi-supervised learning method; analyzing the inductive-deductive two-step path; a method for analyzing and selecting a proper sample without identification and marking; analyzing an anti-interference strategy of semi-supervised learning; analyzing rough set theory and regression model popularity methods;
and a third step of: equipment map state change detection technology analysis and application
a. Unstructured image data structured description techniques:
aiming at the characteristics of large aerial photographing data volume, serious noise interference and rich information, analyzing a basic sparse coding model of image information; analyzing an image sparse coding algorithm based on an optimized dictionary; analyzing a device image feature description algorithm based on overcomplete sparse representation and multi-resolution image sparse representation; a device atlas classification method that analyzes visual word trees; a structured description method for analyzing and realizing unstructured image data;
b. equipment labeling technology based on machine learning model:
in order to improve the retrieval speed and precision of the aerial data, analyzing a visual information labeling frame based on machine learning and oriented to semantic inter-concept association characteristic mining, and further obtaining a device labeling theory and method combining the relativity between concepts; analyzing a visual information labeling technology based on multi-example multi-semantic concept learning; the analysis better eliminates the ambiguity of visual data by mining the interrelation between concepts, and the analysis improves the accuracy of device classification by efficient device labeling and application technology;
soft change detection techniques that can measure the level of device state change:
in order to obtain the running state of equipment through aerial data, analyzing a soft detection method based on infrared equipment maps, ultraviolet equipment maps and visible equipment maps; according to the characteristics of the equipment map, analyzing an equipment state grading detection technology based on a multi-view and multi-period image fusion technology; analyzing a device state classification method and an application technology based on a neural network; analyzing the automatic statistics and marking technology of the equipment state level;
c. the refining and dividing technology of the equipment fault area comprises the following steps:
the main purpose of the aerial data is to find out failure precursors before failure occurs, realize early warning, and quickly locate a failure area when the failure occurs, and realize quick overhaul and recovery; to achieve this objective, the present project proposes a method for analyzing an image based on device labeling information to automatically retrieve a faulty device; analyzing a multi-point image information fusion control technology and analyzing a method for finely dividing images and classifying faults according to equipment map features;
the abstract process of building the database includes real world, information world and machine world data models; the real world includes things and their connections, the information world includes conceptual models, and the machine world data models include relational models, mesh models, and hierarchical models.
The software architecture comprises an application layer, an analysis layer, an integration layer, an access layer and a linkage control interface, wherein the access layer comprises a video camera, a camera, infrared equipment and ultraviolet equipment; the integrated layer comprises comprehensive data acquisition; the analysis layer comprises an appearance parameter, a core algorithm and a switching model; the application layer comprises power transmission line fault identification and early warning processing, power transmission line intelligent inspection state evaluation and statistical analysis management control.
And establishing a learning model principle to learn and infer characteristic extraction and selection of a typical scene database so as to establish a mathematical model.
The invention has the beneficial effects that: (1) The method comprises the steps that an image feature description algorithm is provided through video images acquired through unmanned aerial vehicle inspection based on sparse coding, visual words, feature appearance extraction and other technologies, and structured description of unstructured image data is achieved; providing an object labeling technology based on a machine learning model under a target visual mechanism, and realizing an object recognition and classification method; providing a soft change detection technology, realizing the detection of the state change of the equipment, and dividing the level of the state change of the equipment; the method comprises the steps of providing a refined segmentation technology of equipment fault areas, realizing accurate extraction of the fault areas, and realizing prejudgment and alarm of equipment states with target tasks in a business scene of a company transmission line through the technology;
(2) Based on an aerial photographing data acquisition platform, automatic detection and recognition of equipment map state change in aerial photographing data acquisition are realized, and the efficiency and accuracy of equipment state recognition are improved;
(3) According to the needs of supervised, semi-supervised and weakly supervised machine learning models, a typical scene video/image database (comprising a positive sample library, a negative sample library and the like) is established, a machine learning model such as a cascade classification model, a probability map model, a deep learning model and the like is established, a corresponding parameter learning method and a model-based inference method are provided, and theoretical support is formed for machine learning model modeling and application in multimedia recognition and stream processing;
drawings
FIG. 1 is an abstract process representation of building a database;
FIG. 2 is a system frame diagram of the present invention;
FIG. 3 is a schematic diagram of a learning model building in accordance with the present invention;
description of the embodiments
Embodiment 1, referring to fig. 1, an application of an image recognition algorithm in an intelligent inspection method of an unmanned aerial vehicle for an electric transmission line, and the purpose of the invention is as follows: with the development of electronic information technology, a computer vision analysis technology with artificial intelligence significance is introduced, so that the existing inspection system has the recognition capability of human eyes and the analysis capability of human brains, and potential threat events such as foreign matter hanging, icing, hardware missing, insulation device abnormality and the like on a power transmission line are monitored and analyzed through vision analysis, so that key equipment and environments in the power transmission line are effectively monitored, the reliability of the equipment is early-warned in real time, the state maintenance of related equipment of a power distribution network and the management level of a demand side are further enhanced, the utilization rate of assets is improved, and the accident rate is reduced; therefore, transmission line state detection based on intelligent video analysis has become the most important subject of the current intelligent strong power grid;
according to the principle of three elements of efficiency, quality and cost, an intelligent inspection system based on objective vision is developed and designed, the actual condition and the requirement of the operation inspection management of the power transmission network in China are technically met, the intelligent inspection system is a major breakthrough of the intelligent inspection technology on the intelligent power network level, the inspection efficiency and quality of the power transmission line are greatly improved, the pressure of visual inspection analysis of personnel is reduced, and remarkable social benefit and popularization and application value are achieved;
the analysis application of the intelligent inspection technology of the unmanned aerial vehicle for the power transmission line based on the image recognition algorithm belongs to the front edge technology, after project implementation, the automatic detection and recognition of the state change of the equipment map in the aerial data acquisition are realized, the line operation management level is greatly improved, a new idea is developed for line inspection and state maintenance, and huge economic benefits and social benefits are provided;
according to the invention, through analyzing a preprocessing method of video images, a multimedia recognition and stream processing machine learning model and a device map state change detection technology, a power transmission line unmanned aerial vehicle intelligent inspection system software product based on an image recognition algorithm is researched and developed, the pre-judgment and alarm of the device state with a target task are realized, the automatic detection and recognition of the device map state change in aerial photographing data acquisition are realized, and the efficiency and accuracy of device state recognition are improved;
(1) Preprocessing method analysis of video image
The preprocessing method for the inspection image is analyzed, and a preprocessing algorithm with strong universality, good processing effect and wide application range is found, particularly in the aspects of brightness balance, contrast enhancement, image denoising, deblurring and the like; for the anomaly identification of the aerial transmission line inspection image sequence, key frame extraction is also needed, so that the image information operand is reduced, and the identification accuracy is improved;
at present, a mature inspection image preprocessing method exists, but because helicopter inspection is performed in a field natural environment, image acquisition is easily affected by noise and motion, and noise and motion blurring processing is a problem which needs to be considered in the inspection image preprocessing; in addition to additive noise caused by an optical system and electronic devices, main noise in a helicopter inspection image is multiplicative noise caused by illumination conditions, atmospheric turbulence and the like; homomorphic filtering can be used for eliminating multiplicative noise, and can also be used for denoising images by a wavelet method and an independent component analysis method; after estimating the standard deviation of the image noise, corresponding measures are adopted to improve the peak signal-to-noise ratio of the image so as to achieve a better denoising effect;
when an image is acquired, a helicopter generally moves, and the influence of image blurring caused by the movement on an aerial image is very serious; in order to remove the blurring caused by the motion, the speed and the direction of the motion need to be determined, and as the speed and the direction are difficult to estimate accurately, the conventional deblurring method such as a wiener filtering method or a constrained least squares filtering method is difficult to reach the deblurring requirement, but the existing deblurring algorithm for self-correcting the motion speed and the motion direction has good application effect;
(2) Multimedia recognition and stream processing machine learning model analysis
Analyzing and establishing a typical scene learning database required by supervised, semi-supervised and weakly supervised machine learning models;
an analysis database comprising an analysis of database model building methods; collecting original data; the functions of adding and deleting data in the database are realized; analyzing digital feature extraction and understanding of the image data; analysis of sample labeling methods, and the like;
analyzing and establishing machine learning models such as a cascade classification model, a probability map model, a deep learning model and the like, and verifying;
in the cascade classification model, analyzing cascade strategies of each base learner; analyzing a learning strategy and an reasoning algorithm of the base learner; a confidence threshold setting method of an analysis base learner; verifying and adjusting strategies of the analysis model;
in the establishment of the probability map model, analyzing the relation among random variables; analyzing the directed graph or the undirected graph, and generating a model or a selection method of a discrimination model; verifying and adjusting strategies of the analysis model;
analyzing the deep learning model, comprising: analyzing the hierarchy and mechanism of learners in deep learning to determine model structure and complexity; a method for analyzing the memory consumption of the system and reducing the complexity of the algorithm; verifying and adjusting strategies of the analysis model;
analyzing and verifying a learning method of machine learning model parameters;
a method for establishing an analysis objective function expression; an equation or inequality abstract expression method for analyzing constraint conditions; the convergence and the convergence speed of a parameter optimization algorithm in actual application are inspected;
analyzing and verifying an inference method based on a machine learning model;
performing analysis of a supervised learning method: a marking method for analyzing training samples required by the supervised learning method; analyzing and selecting reasonable sample quantity; analyzing an convincing inferred result evaluation method; specific analysis of popular and effective methods such as decision trees, boosting and Bagging algorithms, artificial neural networks, support vector machines and the like;
performing an analysis of an unsupervised learning method; a method of analyzing a total data distribution inferred from sample information; algorithms such as main analysis cluster analysis, association rule analysis and the like;
performing analysis of a semi-supervised learning method; analyzing the inductive-deductive two-step path; a method for analyzing and selecting a proper sample without identification and marking; analyzing an anti-interference strategy of semi-supervised learning; analyzing popular methods such as rough set theory, regression model and the like;
(3) Equipment map state change detection technology analysis and application
a. Unstructured image data structured description techniques:
aiming at the characteristics of large aerial photographing data volume, serious noise interference, rich information and the like, a basic sparse coding model of image information is analyzed; analyzing an image sparse coding algorithm based on an optimized dictionary; analyzing a device image feature description algorithm based on overcomplete sparse representation and multi-resolution image sparse representation; a device atlas classification method that analyzes visual word trees; a structured description method for analyzing and realizing unstructured image data;
b. equipment labeling technology based on machine learning model:
in order to improve the retrieval speed and precision of the aerial data, analyzing a visual information labeling frame based on machine learning and oriented to semantic inter-concept association characteristic mining, and further obtaining a device labeling theory and method combining the relativity between concepts; analyzing a visual information labeling technology based on multi-example multi-semantic concept learning; the analysis better eliminates the ambiguity of visual data by mining the interrelation between concepts, and the analysis improves the accuracy of device classification by efficient device labeling and application technology;
soft change detection techniques that can measure the level of change in device state:
in order to obtain the running state of equipment through aerial data, analyzing a soft detection method based on infrared equipment maps, ultraviolet equipment maps and visible equipment maps; according to the characteristics of the equipment map, analyzing an equipment state grading detection technology based on a multi-view and multi-period image fusion technology; analyzing a device state classification method and an application technology based on a neural network; analyzing the automatic statistics and marking technology of the equipment state level;
c. the refining and dividing technology of the equipment fault area comprises the following steps:
the main purpose of the aerial data is to find out failure precursors before failure occurs, realize early warning, and quickly locate a failure area when the failure occurs, and realize quick overhaul and recovery; to achieve this objective, the present project proposes a method for analyzing an image based on device labeling information to automatically retrieve a faulty device; analyzing a multi-point image information fusion control technology and analyzing a method for finely dividing images and classifying faults according to equipment map features;
database technology is a core technology of information systems; is a computer-aided data management method that analyzes how data is organized and stored, and how data is efficiently acquired and processed; the objects analyzed and managed by the database technology are data, so the specific contents related to the database technology mainly comprise: through unified organization and management of data, a corresponding database and a data warehouse are established according to a designated structure; the data management and data mining application system capable of achieving various functions of adding, modifying, deleting, processing, analyzing, understanding, reporting, printing and the like on the data in the database is designed by utilizing the database management system and the data mining system; and finally, processing, analyzing and understanding the data by using an application management system;
an intelligent visual analysis processing system framework based on an aerial photographing data acquisition platform of a power transmission line is established by adopting advanced computer technology and automation technology and oriented to a smart power grid and a strong power grid, key technologies in the intelligent visual analysis processing system framework are analyzed, typical application is realized, and a foundation is laid for further realizing analysis, processing and application of massive video data; the technical route of the project surrounds how to develop the analysis of key technologies, and then a set of practical automatic detection system suitable for intelligent power grid equipment is built to develop division work and cooperation, and the analysis is completed in steps and division items to realize the aim;
establishing a cascade classification model, a probability map model and a deep learning model, and carrying out optimization selection on corresponding model parameters: selecting a representation for the learner, i.e., selecting a particular set of classifiers; in the cascade classification model, a proper strategy is selected in different links to find a group of base learners so that the base learners can complement each other.
Claims (4)
1. An application of an image recognition algorithm in an intelligent inspection method of an electric transmission line unmanned aerial vehicle; the method is characterized in that: the first step: preprocessing method analysis of video image
The preprocessing method for the inspection image is analyzed, and a preprocessing algorithm with strong universality, good processing effect and wide application range is found, particularly in the aspects of brightness balance, contrast enhancement, image denoising and deblurring; for the anomaly identification of the aerial transmission line inspection image sequence, key frame extraction is also needed, so that the image information operand is reduced, and the identification accuracy is improved;
and a second step of: multimedia recognition and stream processing machine learning model analysis
Analyzing and establishing a typical scene learning database required by supervised, semi-supervised and weakly supervised machine learning models; an analysis database comprising an analysis of database model building methods; collecting original data; the functions of adding and deleting data in the database are realized; analyzing digital feature extraction and understanding of the image data; analyzing a sample marking method;
analyzing and establishing a cascade classification model, a probability map model and a deep learning model machine learning model, and verifying;
in the cascade classification model, analyzing cascade strategies of each base learner; analyzing a learning strategy and an reasoning algorithm of the base learner; a confidence threshold setting method of an analysis base learner; verifying and adjusting strategies of the analysis model;
in the establishment of the probability map model, analyzing the relation among random variables; analyzing the directed graph or the undirected graph, and generating a model or a selection method of a discrimination model; verifying and adjusting strategies of the analysis model;
analyzing the deep learning model, comprising: analyzing the hierarchy and mechanism of learners in deep learning to determine model structure and complexity; a method for analyzing the memory consumption of the system and reducing the complexity of the algorithm; verifying and adjusting strategies of the analysis model;
analyzing and verifying a learning method of machine learning model parameters;
a method for establishing an analysis objective function expression; an equation or inequality abstract expression method for analyzing constraint conditions; the convergence and the convergence speed of a parameter optimization algorithm in actual application are inspected;
analyzing and verifying an inference method based on a machine learning model;
performing analysis of a supervised learning method: a marking method for analyzing training samples required by the supervised learning method; analyzing and selecting reasonable sample quantity; analyzing an convincing inferred result evaluation method; specifically analyzing decision trees, boosting and Bagging algorithms, artificial neural networks and popular and effective methods of support vector machines;
performing an analysis of an unsupervised learning method; a method of analyzing a total data distribution inferred from sample information; a main analysis cluster analysis and association rule analysis algorithm;
performing analysis of a semi-supervised learning method; analyzing the inductive-deductive two-step path; a method for analyzing and selecting a proper sample without identification and marking; analyzing an anti-interference strategy of semi-supervised learning; analyzing rough set theory and regression model popularity methods;
and a third step of: equipment map state change detection technology analysis and application
a. Unstructured image data structured description techniques:
aiming at the characteristics of large aerial photographing data volume, serious noise interference and rich information, analyzing a basic sparse coding model of image information; analyzing an image sparse coding algorithm based on an optimized dictionary; analyzing a device image feature description algorithm based on overcomplete sparse representation and multi-resolution image sparse representation; a device atlas classification method that analyzes visual word trees; a structured description method for analyzing and realizing unstructured image data;
b. equipment labeling technology based on machine learning model:
in order to improve the retrieval speed and precision of the aerial data, analyzing a visual information labeling frame based on machine learning and oriented to semantic inter-concept association characteristic mining, and further obtaining a device labeling theory and method combining the relativity between concepts; analyzing a visual information labeling technology based on multi-example multi-semantic concept learning; the analysis better eliminates the ambiguity of visual data by mining the interrelation between concepts, and the analysis improves the accuracy of device classification by efficient device labeling and application technology;
soft change detection techniques that can measure the level of device state change:
in order to obtain the running state of equipment through aerial data, analyzing a soft detection method based on infrared equipment maps, ultraviolet equipment maps and visible equipment maps; according to the characteristics of the equipment map, analyzing an equipment state grading detection technology based on a multi-view and multi-period image fusion technology; analyzing a device state classification method and an application technology based on a neural network; analyzing the automatic statistics and marking technology of the equipment state level;
c. the refining and dividing technology of the equipment fault area comprises the following steps:
the main purpose of the aerial data is to find out failure precursors before failure occurs, realize early warning, and quickly locate a failure area when the failure occurs, and realize quick overhaul and recovery; to achieve this objective, the present project proposes a method for analyzing an image based on device labeling information to automatically retrieve a faulty device; a method for analyzing multi-point image information fusion control technology and analyzing sub-divided images and classified faults according to equipment map features.
2. The application of an image recognition algorithm to an intelligent inspection method of an electric power transmission line unmanned aerial vehicle according to claim 1, wherein the abstract process of creating a database comprises real world, information world and machine world data models; the real world includes things and their connections, the information world includes conceptual models, and the machine world data models include relational models, mesh models, and hierarchical models.
3. The application of the image recognition algorithm to the intelligent inspection method of the unmanned aerial vehicle of the power transmission line according to claim 1, wherein the software architecture comprises an application layer, an analysis layer, an integration layer, an access layer and a linkage control interface, and the access layer comprises a video camera, a camera, infrared equipment and ultraviolet equipment; the integrated layer comprises comprehensive data acquisition; the analysis layer comprises an appearance parameter, a core algorithm and a switching model; the application layer comprises power transmission line fault identification and early warning processing, power transmission line intelligent inspection state evaluation and statistical analysis management control.
4. The application of the image recognition algorithm to the intelligent inspection method of the unmanned aerial vehicle of the power transmission line according to claim 1 is characterized in that a learning model is built to learn and infer characteristic extraction and selection of a typical scene database to build a mathematical model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310500114.1A CN116486240A (en) | 2023-05-06 | 2023-05-06 | Application of image recognition algorithm in intelligent inspection method of unmanned aerial vehicle of power transmission line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310500114.1A CN116486240A (en) | 2023-05-06 | 2023-05-06 | Application of image recognition algorithm in intelligent inspection method of unmanned aerial vehicle of power transmission line |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116486240A true CN116486240A (en) | 2023-07-25 |
Family
ID=87223100
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310500114.1A Pending CN116486240A (en) | 2023-05-06 | 2023-05-06 | Application of image recognition algorithm in intelligent inspection method of unmanned aerial vehicle of power transmission line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116486240A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116723412A (en) * | 2023-08-10 | 2023-09-08 | 四川玉米星球科技有限公司 | Method for homogenizing background light and shadow in photo and text shooting and scanning system |
| CN117635905A (en) * | 2023-12-13 | 2024-03-01 | 国网上海市电力公司 | Intelligent monitoring method for electric energy meter attachment quality based on image recognition algorithm |
-
2023
- 2023-05-06 CN CN202310500114.1A patent/CN116486240A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116723412A (en) * | 2023-08-10 | 2023-09-08 | 四川玉米星球科技有限公司 | Method for homogenizing background light and shadow in photo and text shooting and scanning system |
| CN116723412B (en) * | 2023-08-10 | 2023-11-10 | 四川玉米星球科技有限公司 | Method for homogenizing background light and shadow in photo and text shooting and scanning system |
| CN117635905A (en) * | 2023-12-13 | 2024-03-01 | 国网上海市电力公司 | Intelligent monitoring method for electric energy meter attachment quality based on image recognition algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116486240A (en) | Application of image recognition algorithm in intelligent inspection method of unmanned aerial vehicle of power transmission line | |
| CN109741328A (en) | A kind of automobile apparent mass detection method based on production confrontation network | |
| CN110232380A (en) | Fire night scenes restored method based on Mask R-CNN neural network | |
| CN108509954A (en) | A kind of more car plate dynamic identifying methods of real-time traffic scene | |
| CN106408015A (en) | Road fork identification and depth estimation method based on convolutional neural network | |
| CN111444939A (en) | Small-scale equipment component detection method based on weak supervision cooperative learning in open scene of power field | |
| CN109993138A (en) | A kind of car plate detection and recognition methods and device | |
| CN107247952B (en) | Deep supervision-based visual saliency detection method for cyclic convolution neural network | |
| Nazerdeylami et al. | Autonomous litter surveying and human activity monitoring for governance intelligence in coastal eco-cyber-physical systems | |
| CN110599459A (en) | Underground pipe network risk assessment cloud system based on deep learning | |
| CN111079645A (en) | Insulator self-explosion identification method based on AlexNet network | |
| Wang et al. | Railway insulator detection based on adaptive cascaded convolutional neural network | |
| CN111507249A (en) | Transformer substation nest identification method based on target detection | |
| CN117437382B (en) | Updating method and system for data center component | |
| CN112907734A (en) | TEDS fault detection method based on virtual CRH380A model and deep learning | |
| CN117636454A (en) | Intelligent video behavior analysis method based on computer vision | |
| CN116862867A (en) | Small sample transformer substation equipment visual defect detection method and system based on improved AnoGAN | |
| CN116630683A (en) | Road damage detection method based on diffusion self-adaptive feature extraction | |
| Gupta et al. | Post disaster mapping with semantic change detection in satellite imagery | |
| CN115100546A (en) | Mobile-based small target defect identification method and system for power equipment | |
| Ning et al. | Object detection and danger warning of transmission channel based on improved YOLO network | |
| CN114332144A (en) | Sample granularity detection method and system, electronic equipment and storage medium | |
| Lu et al. | Research on target defect detection based on machine vision | |
| Liu et al. | Peaks fusion assisted early-stopping strategy for overhead imagery segmentation with noisy labels | |
| Zhang et al. | Intelligent Detection Model for Power Grids Based on Graph Neural Networks |
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 |