CN112633535A - Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle image - Google Patents
Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle image Download PDFInfo
- Publication number
- CN112633535A CN112633535A CN202110046360.5A CN202110046360A CN112633535A CN 112633535 A CN112633535 A CN 112633535A CN 202110046360 A CN202110046360 A CN 202110046360A CN 112633535 A CN112633535 A CN 112633535A
- Authority
- CN
- China
- Prior art keywords
- image
- inspection
- photovoltaic
- power station
- photovoltaic power
- 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
- 238000007689 inspection Methods 0.000 title claims abstract description 166
- 238000000034 method Methods 0.000 title claims abstract description 59
- 230000007547 defect Effects 0.000 claims abstract description 112
- 238000001514 detection method Methods 0.000 claims abstract description 38
- 230000008569 process Effects 0.000 claims abstract description 26
- 238000013135 deep learning Methods 0.000 claims abstract description 23
- 238000007405 data analysis Methods 0.000 claims abstract description 15
- 230000008676 import Effects 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims description 17
- 238000003062 neural network model Methods 0.000 claims description 15
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 5
- 238000013507 mapping Methods 0.000 claims description 5
- 230000007613 environmental effect Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 230000000712 assembly Effects 0.000 claims description 3
- 238000000429 assembly Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000009877 rendering Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 240000004050 Pentaglottis sempervirens Species 0.000 claims description 2
- 235000004522 Pentaglottis sempervirens Nutrition 0.000 claims description 2
- 238000012423 maintenance Methods 0.000 abstract description 12
- 238000010276 construction Methods 0.000 abstract description 7
- 230000006870 function Effects 0.000 abstract description 7
- 230000000007 visual effect Effects 0.000 abstract description 5
- 238000007726 management method Methods 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 5
- 238000010248 power generation Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 230000002950 deficient Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 210000003608 fece Anatomy 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000032683 aging Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000007664 blowing Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- 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
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- 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
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C1/00—Registering, indicating or recording the time of events or elapsed time, e.g. time-recorders for work people
- G07C1/20—Checking timed patrols, e.g. of watchman
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
-
- 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
- G06T2207/10004—Still image; Photographic image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention discloses a photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle images, wherein the system comprises an unmanned aerial vehicle data import module, an inspection data analysis module, a deep learning algorithm module, a digital photovoltaic power station module, an inspection result management module and an inspection playback module; according to the method, a deep learning algorithm is introduced into the multi-mode patrol image data, so that more accurate photovoltaic defect detection is realized; the method comprises the steps that a digital photovoltaic power station is established, so that the expression of the overall layout of the photovoltaic power station and the playback of a routing inspection process are realized; the positions and the numbers of the photovoltaic panels in the real scene are acquired through registration of the virtual scene image and the real collected image; the single-batch inspection process and result are visually displayed through a playback function, and the defect of the photovoltaic panel is accurately detected; the construction patrols and examines the playback function, and is visual through the process of patrolling and examining, and the management is patrolled and examined to the audio-visual debugging of the operation and maintenance personnel of being convenient for, promotes photovoltaic power plant's automatic operation and maintenance level by a wide margin, effectively improves photovoltaic power plant's the efficiency of patrolling and examining.
Description
Technical Field
The invention relates to a photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle images.
Background
Photovoltaic power generation is an industry that directly converts solar energy into electrical energy. Due to the defects of limited reserves, high pollution and the like of conventional energy sources, the environment-friendly, reliable and renewable photovoltaic power generation system is rapidly developed. The photovoltaic power generation system based on the solar energy is divided into a centralized type and a distributed type, the centralized photovoltaic power station generally occupies a wide area, most of the construction positions are located in some remote areas, and the natural environment is severe; distributed power plants are typically built on roofs, greenhouses and large area pools. In the operation of a photovoltaic power station, a photovoltaic panel is easily influenced by factors such as bird droppings pollution, tree branch and leaf shielding, wind-blowing sun-drying aging and the like, and the problems of cracks, hot spots and the like can occur along with the lapse of time, so that the power generation efficiency of the photovoltaic panel is influenced. This has brought huge fortune dimension pressure to the managers of photovoltaic power plant, needs regularly to patrol photovoltaic power plant in order to check whether there is the defect in the photovoltaic board.
Traditional photovoltaic power plant operation and maintenance mainly relies on the manual work to patrol and examine, and this kind of mode is with high costs not only, work efficiency is low, still need rely on operation and maintenance personnel's experience to differentiate, appears easily patrolling and examining not in place scheduling problem. In addition, in remote areas, the traffic environment is relatively severe, so that the inspection work cost of the photovoltaic power station is high, and the efficiency is low. Therefore, the manual inspection mode is difficult to meet the inspection requirements of safe and efficient photovoltaic power stations.
Along with the development of unmanned aerial vehicle technique and the promotion of national smart power grids construction, provided the photovoltaic power station intelligence system of patrolling and examining based on visual image, it can reduce the required number of people of fortune dimension and time by a wide margin, uses manpower sparingly fortune dimension cost, effectively promotes the power station and patrols and examines efficiency. In some existing intelligent inspection systems for photovoltaic power stations, functions such as routing inspection path planning, intelligent flight of unmanned aerial vehicles, remote image acquisition and transmission, intelligent diagnosis of equipment faults, inspection report generation and the like are generally included, and the conversion from "handheld equipment inspection" to "automatic inspection by unmanned aerial vehicles" has been realized. However, how to sufficiently combine the intelligent technology and further improve the automation level and the refinement degree of the routing inspection of the photovoltaic power station becomes the development direction of the next stage.
In an existing automatic inspection system of a photovoltaic power station, an unmanned aerial vehicle photovoltaic power station detection method disclosed in the publication No. CN107356339A adopts an inspection worker to manually operate an unmanned aerial vehicle to fly in the power station to finish inspection; the method for intelligently inspecting the photovoltaic power station by applying the unmanned aerial vehicle and the inspection device thereof, which are disclosed by the publication number CN110276851A, are used for adhering a coded label to a photovoltaic panel to finish the accurate positioning of a defective photovoltaic panel; the unmanned aerial vehicle and inspection method for automatic inspection of large-scale centralized photovoltaic power stations, which is disclosed in publication number CN111459190A, obtains the approximate coordinates of the defective photovoltaic panel by using longitude and latitude information contained in an exchangeable image file (EXIF). For the photovoltaic power station scene with complex terrain, the inspection methods either still need a large amount of manual operation or are difficult to accurately position the photovoltaic panel assembly, namely the defect detection capability and the automation degree of the photovoltaic power station are still low, and the practicability of the inspection methods in the actual scene is greatly limited.
Disclosure of Invention
The invention aims to provide an intelligent photovoltaic power station inspection method and system based on unmanned aerial vehicle images.
In order to achieve the purpose, the technical scheme of the invention is as follows:
the utility model provides a photovoltaic power plant intelligence inspection method based on unmanned aerial vehicle image, includes that unmanned aerial vehicle flies according to appointed orbit of patrolling and examining and acquires the photovoltaic board image, carries out analytic processing output photovoltaic board fault location to the photovoltaic board image that acquires, wherein, carry out analytic processing output photovoltaic board fault location to the photovoltaic board image that acquires includes:
firstly, establishing a digital photovoltaic power station: respectively generating a virtual simulation scene image and a semantic instance image which are consistent with the actual photovoltaic power station scene image from the actual photovoltaic power station aerial view image; the virtual simulation scene image and the semantic instance image are matched with each other, the virtual simulation scene image comprises a photovoltaic panel model and a scene layout, the scene layout comprises photovoltaic panel model layout and environmental factors, and the photovoltaic panel model layout contains positioning coordinate information based on geographical longitude and latitude; the semantic instance image is obtained by distinguishing and expressing different photovoltaic panel models of the layout by using rendering images and numbers;
secondly, pixel extraction is carried out on the obtained photovoltaic panel image, the extracted pixels are input into a defect judgment neural network model, and a judgment result is output by the defect judgment neural network model, wherein: the output of the defect result comprises defect type information and coordinates of a defect bounding box;
mapping the output defect type information and the defect boundary frame coordinates to the semantic case image, acquiring the serial number information of the photovoltaic panel where the defect is located from the semantic case image, and displaying the defect boundary frame in a virtual simulation scene image photovoltaic panel model to position the fault of the photovoltaic panel according to the mutual matching relation between the virtual simulation scene image and the semantic case image;
storing inspection result data for inspection playback inquiry and outputting final photovoltaic panel fault location;
wherein: the defect judgment neural network model is a model which is built by deep learning in advance according to the defect type of the photovoltaic panel.
The scheme is further as follows: the photovoltaic panel image comprises a visible light image and an infrared light image which are continuously photographed at intervals, the visible light image is used for judging whether sundries are sheltered on the surface of the photovoltaic panel or not and judging the type of the sundries, and the infrared light image is used for judging whether hot spot blocks exist on the surface of the photovoltaic panel or not, wherein: at least one-third of the interval images overlap.
The scheme is further as follows: the specified track is a track which is made from a virtual simulation scene image according to geographic longitude and latitude positioning coordinate information contained in the photovoltaic panel model layout.
The scheme is further as follows: the inspection result data comprises longitude and latitude and height information of a flight track of the unmanned aerial vehicle, shooting time and defect types of images, and information of defect positions and numbers of photovoltaic panels where the defect positions are located.
The scheme is further as follows: the defect judgment neural network model uses a fast-RCNN target detection model.
The scheme is further as follows: the output final photovoltaic panel fault location is as follows: unmanned aerial vehicle patrols and examines the orbit repeatedly many times according to the appointed, acquires the photovoltaic board image each time and carries out analytic processing and output photovoltaic board fault location respectively, compares photovoltaic board fault location each time, regards the photovoltaic board fault location who appears repeatedly as final photovoltaic board fault location output.
A system for realizing the intelligent inspection method of the photovoltaic power station comprises the following steps: the system comprises an unmanned aerial vehicle data import module, a patrol data analysis module, a deep learning algorithm module, a digital photovoltaic power station module, a patrol result management module and a patrol playback module; wherein:
the unmanned aerial vehicle data import module is used for preprocessing an infrared light image and a visible light image acquired by unmanned aerial vehicle inspection and importing the infrared light image and the visible light image into the inspection data analysis module;
the inspection data analysis module analyzes the inspection image data into a required inspection result by using a deep learning algorithm, wherein the inspection result comprises longitude and latitude, height, shooting time, defect types, defect positions and numbers of photovoltaic panels where the inspection image data are located;
the deep learning algorithm module provides support for the patrol data analysis module and comprises a defect detection algorithm and an image registration algorithm, wherein the defect detection algorithm detects various defects from the multi-modal patrol image, and the image registration algorithm registers the virtual simulation scene image of the digital photovoltaic power station module with the actually acquired real photovoltaic power station image;
the digital photovoltaic power station module establishes a corresponding virtual simulation scene image and a semantic instance image for an actual photovoltaic power station, generates scene image and semantic instance image virtual data for the deep learning algorithm module, and further visualizes the routing inspection process of the routing inspection playback module;
the inspection result management module is used for storing and managing data of the photovoltaic power station inspection, including original inspection data and analyzed inspection results, and supporting the query of the inspection results;
the inspection playback module is used for visualizing inspection process data and result data of a single batch to realize playback display in the digital photovoltaic power station.
The scheme is further as follows: the defect detection algorithm is used for detecting the defects of the photovoltaic panel in the inspection picture by adopting a Faster-RCNN target detection model; the image registration algorithm is characterized in that the feature point positions of two images and corresponding descriptors are matched by using a BFMatcher feature point matching method, unqualified feature matching point pairs are removed according to a random sampling consistency algorithm ransac, and then an affine transformation matrix is calculated by using the remaining matching feature point pairs, so that registration between a real inspection image and a virtual scene image is realized.
The scheme is further as follows: the method for detecting the defects of the photovoltaic panel in the inspection picture by the fast-RCNN target detection model comprises the following steps: firstly, a pre-trained convolutional neural network is used as a feature extractor to generate a convolutional feature map, then a region candidate network RPN is used for extracting a candidate region, and finally feature region Pooling RoI Pooling processing is adopted on the features extracted by the convolutional neural network and a boundary box containing related objects to adjust the coordinates of the boundary box.
The scheme is further as follows: the process of establishing the corresponding virtual simulation scene image and the semantic instance image for the actual photovoltaic power station is as follows: firstly, building a 1:1 virtual model of a real power station scene in a virtual scene by using a bird's-eye view photo of a photovoltaic power station shot by an unmanned aerial vehicle on site, constructing a three-dimensional model of a virtual photovoltaic panel, and adjusting the specific position of each photovoltaic panel in the scene according to the serial number and longitude and latitude coordinate data of the photovoltaic panel to make the layout of the photovoltaic panels in the digital photovoltaic power station consistent with that of an actual power station to obtain a virtual simulation photovoltaic power station scene image; when each virtual scene photovoltaic panel picture is generated in the image, an image generation plug-in is constructed by using a phantom 4 engine, a mask image of a photovoltaic panel example is rendered while each virtual scene image of the photovoltaic panel is generated, different photovoltaic panel assemblies are expressed by adopting corresponding numbers, and finally a virtual simulation scene image and a semantic example image which are matched with a real acquisition image of a photovoltaic power station are generated.
The invention has the beneficial effects that:
according to the method, a deep learning algorithm is introduced into the multi-mode patrol image data, so that more accurate photovoltaic defect detection is realized; the method comprises the steps that a digital photovoltaic power station is established, so that the expression of the overall layout of the photovoltaic power station and the playback of a routing inspection process are realized; the positions and the numbers of the photovoltaic panels in the real scene are acquired through registration of the virtual scene image and the real collected image; and the single-batch inspection process and result are visually displayed through a playback function.
The invention solves the problems of low automation degree, inaccurate detection result and the like of manual inspection, thereby improving the operation and maintenance efficiency of the photovoltaic power station and reducing the operation and maintenance cost. The inspection unmanned aerial vehicle acquires image data with positioning information to support automatic positioning of defects of the photovoltaic power station; establishing a digital photovoltaic power station with instance labels, and realizing automatic positioning and serial number generation of a photovoltaic panel assembly in the inspection image through registration of a virtual scene graph and a real acquisition graph; processing multi-mode data such as visible light, infrared and the like by utilizing a deep learning algorithm to realize accurate detection of defects of the photovoltaic panel; the construction patrols and examines the playback function, and is visual through the process of patrolling and examining, and the management is patrolled and examined to the audio-visual debugging of the operation and maintenance personnel of being convenient for, promotes photovoltaic power plant's automatic operation and maintenance level by a wide margin, effectively improves photovoltaic power plant's the efficiency of patrolling and examining.
The invention is described in detail below with reference to the figures and examples.
Drawings
FIG. 1 is a block diagram of an intelligent inspection system of a photovoltaic power station;
FIG. 2 is a flow chart of the intelligent inspection process of the photovoltaic power station of the present invention;
FIG. 3 is a schematic diagram illustrating a comparison between a real scene and a virtual scene of a photovoltaic power station according to the present invention;
FIG. 4 is a schematic diagram of a comparison of three images according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the description of the present embodiment, it should be noted that the terms "connected" and "disposed" should be interpreted broadly, for example, the "connection" may be a wire connection or a mechanical connection; the 'placing' can be fixedly connected placing or integrally formed placing. The specific meanings of the above terms in the present embodiment can be understood by those of ordinary skill in the art according to specific situations.
The utility model provides a photovoltaic power plant intelligence inspection method based on unmanned aerial vehicle image, includes that unmanned aerial vehicle flies according to appointed orbit of patrolling and examining and acquires the photovoltaic board image, carries out analytic processing output photovoltaic board fault location to the photovoltaic board image that acquires, wherein, carry out analytic processing output photovoltaic board fault location to the photovoltaic board image that acquires includes:
firstly, establishing a digital photovoltaic power station: respectively generating a virtual simulation scene image and a semantic instance image which are consistent with the actual photovoltaic power station scene image from the actual photovoltaic power station aerial view image; the virtual simulation scene image and the semantic instance image are matched with each other, the virtual simulation scene image comprises a photovoltaic panel model and a scene layout, the scene layout comprises the photovoltaic panel model layout and environmental factors (such as roads, trees and the like), and the photovoltaic panel model layout contains positioning coordinate information based on geographical longitude and latitude; the semantic instance image is obtained by distinguishing and expressing different photovoltaic panel models of the layout by using rendering images and numbers;
secondly, pixel extraction is carried out on the obtained photovoltaic panel image, the extracted pixels are input into a defect judgment neural network model, and a judgment result is output by the defect judgment neural network model, wherein: the output of the defect result comprises defect type information and coordinates of a defect bounding box;
mapping the output defect type information and the defect boundary frame coordinates to the semantic case image, acquiring the serial number information of the photovoltaic panel where the defect is located from the semantic case image, and displaying the defect boundary frame in a virtual simulation scene image photovoltaic panel model to position the fault of the photovoltaic panel according to the mutual matching relation between the virtual simulation scene image and the semantic case image;
storing inspection result data for inspection playback inquiry and outputting final photovoltaic panel fault location;
wherein: the defect judgment neural network model is a model which is built by deep learning in advance according to the defect type of the photovoltaic panel.
The photovoltaic panel image comprises a visible light image and an infrared light image which are continuously photographed at intervals, the visible light image is used for judging whether sundries are sheltered on the surface of the photovoltaic panel or not and judging the type of the sundries, and the infrared light image is used for judging whether hot spot blocks exist on the surface of the photovoltaic panel or not, wherein: the interval images are at least one third overlapped, the interval of photographing time can be determined according to the principle of the interval of the flying speed of the unmanned aerial vehicle, or the flying height of the unmanned aerial vehicle is determined according to the interval time, and the defect judgment neural network model comprises a visible light image photovoltaic panel defect type neural network model and an infrared light image photovoltaic panel defect type neural network model; the defect judgment neural network model uses a fast-RCNN target detection model, the fast-RCNN target detection model can automatically extract image features, and can output coordinates of a defect boundary box.
The specified track is a track which is formulated from a virtual simulation scene image according to geographic longitude and latitude positioning coordinate information contained in the photovoltaic panel model layout. The inspection result data comprises longitude and latitude and height information of a flight track of the unmanned aerial vehicle, shooting time and defect types of images, and information of defect positions and numbers of photovoltaic panels where the defect positions are located.
In order to more accurately position the defect position and prevent false defect positioning, the method for positioning the fault of the output final photovoltaic panel by utilizing routing inspection playback inquiry comprises the following steps: unmanned aerial vehicle patrols and examines the orbit repeatedly many times according to the appointed, acquires the photovoltaic board image each time and carries out analytic processing and output photovoltaic board fault location respectively, compares photovoltaic board fault location each time, regards the photovoltaic board fault location who appears repeatedly as final photovoltaic board fault location output.
The method is realized by adopting a modular system, and fig. 1 illustrates a system for realizing the intelligent inspection method of the photovoltaic power station based on the unmanned aerial vehicle image, which comprises an unmanned aerial vehicle data import module, an inspection data analysis module, a deep learning algorithm module, a digital photovoltaic power station module, an inspection result management module and an inspection playback module. Wherein:
1) the unmanned aerial vehicle data import module is used for carrying out standardized preprocessing on an infrared light image and a visible light image acquired by unmanned aerial vehicle inspection, for example, the image is clearer through denoising processing, and pixel information of the image is imported into the inspection data analysis module;
2) the inspection data analysis module analyzes the inspection image data into a required inspection result by using a deep learning algorithm, wherein the inspection result comprises longitude and latitude, height, shooting time, defect types, defect positions, numbers of photovoltaic panels where the inspection image data are located and the like;
3) the deep learning algorithm module provides support for the patrol data analysis module, and comprises a defect detection algorithm and an image registration algorithm, wherein the defect detection algorithm detects various defects from the multi-modal patrol image, and the image registration algorithm registers the virtual simulation scene image of the digital photovoltaic power station module with the actually acquired real photovoltaic power station image;
4) the digital photovoltaic power station module establishes corresponding virtual simulation scene images and semantic instance images for an actual photovoltaic power station, on one hand, virtual data of the scene images and the semantic instance images are generated for the deep learning algorithm module, and on the other hand, the visualization of the routing inspection playback module on the routing inspection process is supported;
5) the inspection result management module is used for storing and managing all data of the photovoltaic power station inspection, including original inspection data and analyzed inspection results, and supporting the inquiry of the inspection results;
6) the inspection playback module is used for visualizing inspection process data and result data of a single batch to realize playback display in the digital photovoltaic power station.
Patrol and examine data analysis module and patrol and examine data to unmanned aerial vehicle and analyze, obtain required result of patrolling and examining, handle the collection image of every position in proper order in the realization, to single position data, analysis is handled and is mainly contained following step:
1) analyzing shooting longitude and latitude, height and shooting time parameters contained in the exchangeable image file, judging whether a shooting place is in a target photovoltaic power station or not by using the longitude and latitude parameters, and if not, needing no subsequent operation;
2) using a defect detection algorithm to detect the defects of the visible light image and the infrared image, if the defects exist, giving the positions of the defects in the inspection image in a rectangular frame or mask mode, and giving the types of the defects; otherwise, skipping to process the next data;
3) acquiring corresponding virtual scene images and semantic instance images in the digital photovoltaic power station according to the shooting longitude and latitude and the shooting height of the actual images;
4) registering the real inspection image and the virtual scene image by using an image registration algorithm, and mapping pixels of the real inspection image into a semantic example image to realize photovoltaic panel example analysis of the real inspection image;
5) and obtaining corresponding semantic instance image information according to the position of the defect detection frame or the mask in the real inspection image, thereby obtaining the serial number of the photovoltaic panel where the defect is located.
The deep learning algorithm module is used for supporting the patrol data analysis module and comprises a defect detection algorithm and an image registration algorithm. The defect detection algorithm adopts a deep learning detection model to detect visible light and infrared multi-mode images, the type and detection frame of each defect are obtained, different mode images express information of the same position, and detection results are output together. The image registration algorithm respectively generates feature descriptions for the real inspection image and the virtual scene image, calculates an affine transformation matrix between the real inspection image and the virtual scene image, and then maps pixels of the real inspection image into a semantic instance image which is completely matched with the virtual scene image by adopting affine transformation.
The digital photovoltaic power station module refers to a virtual three-dimensional virtual scene corresponding to a given photovoltaic power station, and comprises a photovoltaic panel model and scene layout, wherein the scene layout comprises photovoltaic panel layout and environmental factors (such as roads, trees and the like). The module supports virtual data generation at a given location, including visible light images of a scene and semantic instance images representing different photovoltaic panels, where the different photovoltaic panels are expressed by fixed numbers.
The inspection result management module is used for organizing, storing and managing inspection data and inspection results and comprises an inspection database and an inspection result query module. The inspection database mainly stores an inspection information table, a single data record table and a defect result table, and the inspection information table stores inspection batch, date, defect number of corresponding batches and other batch information; the single data record table stores inspection information such as inspection batch, image storage address, height, longitude and latitude; and the defect result table stores defect information such as inspection batch, defect type, defect photovoltaic panel number and the like. And the inspection result query module retrieves and displays the database according to different query conditions, and mainly comprises inspection time query, defect query and the like.
The inspection playback module is used for visually displaying the single inspection process, so that a user can conveniently check and analyze the setting and inspection results of the inspection route. For a given single patrol batch, the playback process essentially comprises the following steps:
1) acquiring real inspection images of the batch to be played back and an analyzed inspection result from an inspection result management module, processing the inspection results according to a time sequence, and generating a flight route of an inspection process according to the flight height and the longitude and latitude;
2) marking the photovoltaic panel number at the center of the photovoltaic panel of the real inspection image according to the defect photovoltaic panel number and the defect type recorded in each inspection result, and representing the defects of the photovoltaic panels of different types by using detection frames with different colors;
3) and simulating the polling flight process in the digital photovoltaic power station according to the set polling flight route and the set flight speed, and displaying the polling image with the analysis result.
As shown in fig. 2, in order to realize intelligent inspection of a photovoltaic power station without depending on manual intervention, an artificial intelligence algorithm is used for analyzing and processing inspection data: firstly, real-scene shooting is carried out in a photovoltaic power station through an unmanned aerial vehicle, and a real inspection image of the photovoltaic power station is obtained; then, on one hand, information such as shooting time, longitude and latitude, height and the like of the image is analyzed, and the defect position and the defect type of the photovoltaic panel in the inspection image are detected by using a defect detection algorithm; and on the other hand, the analyzed longitude and latitude and the analyzed height are sampled in the virtual scene to generate a corresponding virtual scene image and an example image, the image registration module registers the real inspection image and the virtual scene image, and the serial number of the photovoltaic panel where the defect in the real scene is located is obtained according to the registration result. After the analysis processing module analyzes the inspection image, the analysis result is stored in the inspection database, and the required inspection result information is provided for the corresponding module when the user inquires and plays back the inspection result.
Unmanned aerial vehicle patrols and examines data's collection
Unmanned aerial vehicle can accomplish the live-action shooting to photovoltaic power plant, has following characteristic:
1) carrying a pan-tilt camera as a load, wherein the pan-tilt camera can simultaneously output a visible light image and an infrared light image;
2) by combining the modes of visual positioning, Global Positioning System (GPS) positioning, real-time kinematic (RTK) positioning and the like, the unmanned aerial vehicle can accurately position the longitude and latitude and the height of the unmanned aerial vehicle;
3) the unmanned aerial vehicle is provided with two obstacle avoidance systems, namely a vision system and an infrared sensing system, so that the unmanned aerial vehicle is ensured not to be dangerous due to the existence of obstacles in the inspection process;
4) the battery endurance carried by the unmanned aerial vehicle can meet the requirement of routing inspection, and once the electric quantity is too low, the unmanned aerial vehicle can automatically return to ensure the flight safety;
5) the unmanned aerial vehicle adopts a navigation mode combining an auxiliary navigation system and an inertial navigation system, and can ensure that the unmanned aerial vehicle shoots along a set routing inspection route.
According to whether manual intervention operation is needed in the inspection process, the inspection mode is divided into a manual mode and a GPS cruise mode. In order to realize automatic routing inspection, the embodiment adopts a GPS cruise mode and comprises the following steps:
1) a scene map is input into the unmanned aerial vehicle by using an unmanned aerial vehicle remote controller, a patrol route is planned in the map by calibrating a navigation point (namely a point where a route turns), and the patrol route is input into the unmanned aerial vehicle for storage. Wherein the planned path should be able to cover all photovoltaic panels in the scene.
2) And setting parameters such as the flying height, the photographing interval, the orientation of the cradle head, the return point and the like of the unmanned aerial vehicle. The set flying height can ensure that the unmanned aerial vehicle is not influenced by the barrier and can clearly shoot the inspection image; the set shooting interval is related to the flying height, and the information of the photovoltaic panel can be ensured not to be missed by the front picture and the rear picture; the direction of the cradle head is always vertical and downward; the set return point of the unmanned aerial vehicle can guarantee the return safety of the unmanned aerial vehicle and meet the endurance limit.
3) And charging the unmanned aerial vehicle, installing an undercarriage, a holder camera and other preparation works, and finally starting the power supply of the unmanned aerial vehicle to enable the unmanned aerial vehicle to shoot the live-action scene of the power station.
In the examples: the process of establishing the corresponding virtual simulation scene image and the semantic instance image for the actual photovoltaic power station is as follows: firstly, a photovoltaic power station overlooking aerial view picture shot by an unmanned aerial vehicle on site is used, and a virtual model 1:1 of a real power station scene is built in a virtual scene by using a phantom 4 engine. Objects to be constructed in a scene are classified according to material attributes, such as land, roads, trees, transformer boxes and the like, and different construction methods are adopted for the objects with different attributes. The method comprises the following steps of carrying out fine construction and reduction on objects with small quantity and low repeatability, such as roads, transformer boxes and the like, so that each object has a 1:1 reduced virtual three-dimensional model; constructing a plurality of random virtual three-dimensional models for objects with large quantity and randomness, such as land, trees and the like, and randomly generating the objects at the corresponding positions of the scene; for a core component of a photovoltaic panel, firstly, according to a top view of a photovoltaic power station and parameters of the photovoltaic panel, such as length, width, inclination angle and distance, obtained through field measurement, a three-dimensional model of the virtual photovoltaic panel is accurately constructed through steps of vertex creation, triangle drawing, color mapping drawing, model material creation, light shadow adding and the like. After the photovoltaic panel is generated, certain random noise textures and ambient light are added to the surface of each photovoltaic panel, so that color change of the photovoltaic panel is realized.
And combining the photovoltaic panel numbers and the longitude and latitude coordinate data in the multiple real inspection pictures, and finely adjusting the specific position of each photovoltaic panel in the scene to make the layout of the photovoltaic panels in the digital photovoltaic power station approximately consistent with that of the actual power station. The obtained virtual simulation photovoltaic power station scene image is shown in fig. 3, the left view is a shot photovoltaic power station live view, and the right view is a generated virtual simulation photovoltaic power station scene.
When each virtual scene picture is generated, an image generation plug-in is constructed by using a phantom 4 engine, a mask image of a photovoltaic panel example is rendered while the virtual scene image of each photovoltaic power station is generated, different photovoltaic panel assemblies are expressed by adopting corresponding numbers, and finally a virtual scene image and a semantic example image which are matched with a real acquired image of the photovoltaic power station are generated, as shown in fig. 4.
Aiming at the common defect types in the photovoltaic power station, the defect detection algorithm of the embodiment adopts a Faster-RCNN target detection model to detect the defects of the photovoltaic panel in the inspection picture. Specifically, firstly, a pre-trained convolutional neural network is used as a feature extractor to generate a convolutional feature map, then a region candidate network (RPN) is used for extracting a candidate region, and finally feature region Pooling (RoI Pooling) is adopted for the features extracted by the convolutional neural network and a bounding box containing a related object to adjust the coordinates of the bounding box.
The embodiment respectively trains a defect detection model according to two modal data of a visible light image and an infrared image. Specifically, the visible light image is mainly used for detecting defects such as bird droppings, tree branch blocking and the like, and the infrared image is mainly used for detecting hot spots. In the training stage, a training set and a test set are respectively divided for images of two modes, and corresponding models are respectively trained; and in the testing stage, fusing the defect detection results output by the two models to be finally output.
In the embodiment, aiming at the registration between the real patrol inspection image and the virtual scene image, firstly, a basic data set formed by images with obvious characteristics such as line segments, triangles, rectangles and the like is used for training a basic model of a backbone characteristic extraction network; secondly, performing multiple homography transformations on each image to perform data enhancement, retraining the labels predicted by the basic model on the real images and the virtual images again, and taking the image feature points predicted by the retrained basic model as feature point labels of each image; then, inputting the original image into a decoding network through coding, and extracting a feature point heat map and a corresponding feature descriptor by using two branches respectively; and finally, matching the feature point positions of the two images with the corresponding descriptors by using a BFMatcher feature point matching method, removing unqualified feature matching point pairs according to a random sampling consensus algorithm (ransac), and calculating an affine transformation matrix by using the remaining matched feature point pairs, thereby realizing the registration between the real inspection image and the virtual scene image.
According to the embodiment, the deep learning algorithm is introduced into the multi-mode patrol inspection image data, so that more accurate photovoltaic defect detection is realized; the method comprises the steps that a digital photovoltaic power station is established, so that the expression of the overall layout of the photovoltaic power station and the playback of a routing inspection process are realized; the positions and the numbers of the photovoltaic panels in the real scene are acquired through registration of the virtual scene image and the real collected image; and the single-batch inspection process and result are visually displayed through a playback function.
The embodiment solves the problems of low automation degree, inaccurate detection result and the like of manual inspection, thereby improving the operation and maintenance efficiency of the photovoltaic power station and reducing the operation and maintenance cost. The inspection unmanned aerial vehicle acquires image data with positioning information to support automatic positioning of defects of the photovoltaic power station; establishing a digital photovoltaic power station with instance labels, and realizing automatic positioning and serial number generation of a photovoltaic panel assembly in the inspection image through registration of a virtual scene graph and a real acquisition graph; processing multi-mode data such as visible light, infrared and the like by utilizing a deep learning algorithm to realize accurate detection of defects of the photovoltaic panel; the construction patrols and examines the playback function, and is visual through the process of patrolling and examining, and the management is patrolled and examined to the audio-visual debugging of the operation and maintenance personnel of being convenient for, promotes photovoltaic power plant's automatic operation and maintenance level by a wide margin, effectively improves photovoltaic power plant's the efficiency of patrolling and examining.
Claims (10)
1. The utility model provides a photovoltaic power plant intelligence inspection method based on unmanned aerial vehicle image, includes that unmanned aerial vehicle flies according to appointed orbit of patrolling and examining and acquires the photovoltaic board image, carries out analytic processing output photovoltaic board fault location to the photovoltaic board image that acquires, its characterized in that, carry out analytic processing output photovoltaic board fault location to the photovoltaic board image that acquires includes:
firstly, establishing a digital photovoltaic power station: respectively generating a virtual simulation scene image and a semantic instance image which are consistent with the actual photovoltaic power station scene image from the actual photovoltaic power station aerial view image; the virtual simulation scene image and the semantic instance image are matched with each other, the virtual simulation scene image comprises a photovoltaic panel model and a scene layout, the scene layout comprises photovoltaic panel model layout and environmental factors, and the photovoltaic panel model layout contains positioning coordinate information based on geographical longitude and latitude; the semantic instance image is obtained by distinguishing and expressing different photovoltaic panel models of the layout by using rendering images and numbers;
secondly, pixel extraction is carried out on the obtained photovoltaic panel image, the extracted pixels are input into a defect judgment neural network model, and a judgment result is output by the defect judgment neural network model, wherein: the output of the defect result comprises defect type information and coordinates of a defect bounding box;
mapping the output defect type information and the defect boundary frame coordinates to the semantic case image, acquiring the serial number information of the photovoltaic panel where the defect is located from the semantic case image, and displaying the defect boundary frame in a virtual simulation scene image photovoltaic panel model to position the fault of the photovoltaic panel according to the mutual matching relation between the virtual simulation scene image and the semantic case image;
storing inspection result data for inspection playback inquiry and outputting final photovoltaic panel fault location;
wherein: the defect judgment neural network model is a model which is built by deep learning in advance according to the defect type of the photovoltaic panel.
2. The method according to claim 1, wherein the photovoltaic panel image comprises a visible light image and an infrared light image which are continuously taken at intervals, the visible light image is used for judging whether the surface of the photovoltaic panel is blocked by sundries and the type of the sundries, and the infrared light image is used for judging whether the surface of the photovoltaic panel has hot spot areas, wherein: at least one-third of the interval images overlap.
3. The method of claim 1, wherein the specified trajectory is a trajectory formulated from a virtual simulation scene image according to geographic latitude and longitude positioning coordinate information contained in a photovoltaic panel model layout.
4. The method according to claim 1, wherein the inspection result data comprises longitude and latitude and height information of a flight path of the unmanned aerial vehicle, shooting time and defect type of the image, and information of a defect position and a photovoltaic panel number where the defect position is located.
5. The method according to claim 1, wherein the deficit-determination neural network model uses a fast-RCNN object detection model.
6. The method of claim 1, wherein the output final photovoltaic panel fault location is: unmanned aerial vehicle patrols and examines the orbit repeatedly many times according to the appointed, acquires the photovoltaic board image each time and carries out analytic processing and output photovoltaic board fault location respectively, compares photovoltaic board fault location each time, regards the photovoltaic board fault location who appears repeatedly as final photovoltaic board fault location output.
7. A system for realizing the intelligent inspection method for the photovoltaic power station, which is characterized by comprising the following steps: the system comprises an unmanned aerial vehicle data import module, a patrol data analysis module, a deep learning algorithm module, a digital photovoltaic power station module, a patrol result management module and a patrol playback module; wherein:
the unmanned aerial vehicle data import module is used for preprocessing an infrared light image and a visible light image acquired by unmanned aerial vehicle inspection and importing the infrared light image and the visible light image into the inspection data analysis module;
the inspection data analysis module analyzes the inspection image data into a required inspection result by using a deep learning algorithm, wherein the inspection result comprises longitude and latitude, height, shooting time, defect types, defect positions and numbers of photovoltaic panels where the inspection image data are located;
the deep learning algorithm module provides support for the patrol data analysis module and comprises a defect detection algorithm and an image registration algorithm, wherein the defect detection algorithm detects various defects from the multi-modal patrol image, and the image registration algorithm registers the virtual simulation scene image of the digital photovoltaic power station module with the actually acquired real photovoltaic power station image;
the digital photovoltaic power station module establishes a corresponding virtual simulation scene image and a semantic instance image for an actual photovoltaic power station, generates scene image and semantic instance image virtual data for the deep learning algorithm module, and further visualizes the routing inspection process of the routing inspection playback module;
the inspection result management module is used for storing and managing data of the photovoltaic power station inspection, including original inspection data and analyzed inspection results, and supporting the query of the inspection results;
the inspection playback module is used for visualizing inspection process data and result data of a single batch to realize playback display in the digital photovoltaic power station.
8. The system according to claim 7, wherein the defect detection algorithm is to detect the defects of the photovoltaic panel in the inspection picture by adopting a fast-RCNN target detection model; the image registration algorithm is characterized in that the feature point positions of two images and corresponding descriptors are matched by using a BFMatcher feature point matching method, unqualified feature matching point pairs are removed according to a random sampling consistency algorithm ransac, and then an affine transformation matrix is calculated by using the remaining matching feature point pairs, so that registration between a real inspection image and a virtual scene image is realized.
9. The system according to claim 8, wherein the Faster-RCNN object detection model for detecting defects of the photovoltaic panel in the inspection picture comprises: firstly, a pre-trained convolutional neural network is used as a feature extractor to generate a convolutional feature map, then a region candidate network RPN is used for extracting a candidate region, and finally feature region Pooling RoI Pooling processing is adopted on the features extracted by the convolutional neural network and a boundary box containing related objects to adjust the coordinates of the boundary box.
10. The system of claim 7, wherein the process of establishing the corresponding virtual simulation scene image and semantic instance image for the actual photovoltaic power plant is: firstly, building a 1:1 virtual model of a real power station scene in a virtual scene by using a bird's-eye view photo of a photovoltaic power station shot by an unmanned aerial vehicle on site, constructing a three-dimensional model of a virtual photovoltaic panel, and adjusting the specific position of each photovoltaic panel in the scene according to the serial number and longitude and latitude coordinate data of the photovoltaic panel to make the layout of the photovoltaic panels in the digital photovoltaic power station consistent with that of an actual power station to obtain a virtual simulation photovoltaic power station scene image; when each virtual scene photovoltaic panel picture is generated in the image, an image generation plug-in is constructed by using a phantom 4 engine, a mask image of a photovoltaic panel example is rendered while each virtual scene image of the photovoltaic panel is generated, different photovoltaic panel assemblies are expressed by adopting corresponding numbers, and finally a virtual simulation scene image and a semantic example image which are matched with a real acquisition image of a photovoltaic power station are generated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110046360.5A CN112633535A (en) | 2021-01-14 | 2021-01-14 | Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110046360.5A CN112633535A (en) | 2021-01-14 | 2021-01-14 | Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112633535A true CN112633535A (en) | 2021-04-09 |
Family
ID=75294159
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110046360.5A Pending CN112633535A (en) | 2021-01-14 | 2021-01-14 | Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112633535A (en) |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113298973A (en) * | 2021-05-26 | 2021-08-24 | 广东电网有限责任公司电力调度控制中心 | Unmanned aerial vehicle power inspection method and device based on 5G communication |
| CN113325877A (en) * | 2021-08-02 | 2021-08-31 | 旻投电力发展有限公司 | RTK positioning-based photovoltaic module positioning method for unmanned aerial vehicle platform |
| CN113325878A (en) * | 2021-08-04 | 2021-08-31 | 旻投电力发展有限公司 | Photovoltaic module dust deposition detection system and detection method based on unmanned aerial vehicle |
| CN113344002A (en) * | 2021-07-29 | 2021-09-03 | 北京图知天下科技有限责任公司 | Target coordinate duplication eliminating method and system, electronic equipment and readable storage medium |
| CN113435369A (en) * | 2021-07-01 | 2021-09-24 | 华风数据(深圳)有限公司 | Photovoltaic fault intelligent identification method based on SSD algorithm model and far infrared video |
| CN113436176A (en) * | 2021-07-01 | 2021-09-24 | 华风数据(深圳)有限公司 | Photovoltaic panel surface fault detection method based on SSD and DSSD algorithm under visible light |
| CN113744270A (en) * | 2021-11-05 | 2021-12-03 | 南京市特种设备安全监督检验研究院 | Unmanned aerial vehicle visual detection and identification method for crane complex steel structure surface defects |
| CN114089786A (en) * | 2021-09-29 | 2022-02-25 | 北京航空航天大学杭州创新研究院 | Autonomous inspection system based on unmanned aerial vehicle vision and along mountain highway |
| CN114285374A (en) * | 2021-12-14 | 2022-04-05 | 上海电气集团股份有限公司 | Equipment positioning method and system for photovoltaic power station, electronic equipment and storage medium |
| CN114898232A (en) * | 2022-04-29 | 2022-08-12 | 中科云尚(南京)智能技术有限公司 | Photovoltaic power station unmanned aerial vehicle inspection method and system based on photovoltaic string data analysis |
| CN114913619A (en) * | 2022-04-08 | 2022-08-16 | 华能苏州热电有限责任公司 | Intelligent mobile inspection method and system |
| CN115049663A (en) * | 2022-08-16 | 2022-09-13 | 星逻智能科技(苏州)有限公司 | Unmanned aerial vehicle-based double-light fusion insulator defect detection method and system |
| CN115202404A (en) * | 2022-09-15 | 2022-10-18 | 广东容祺智能科技有限公司 | Photovoltaic power generation unit maintenance inspection method based on unmanned aerial vehicle |
| CN115239893A (en) * | 2022-09-23 | 2022-10-25 | 运易通科技有限公司 | Image reconstruction method for detecting defects of solar panel of warehouse ceiling |
| CN115767040A (en) * | 2023-01-06 | 2023-03-07 | 松立控股集团股份有限公司 | 360-degree panoramic monitoring automatic cruise method based on interactive continuous learning |
| CN115861855A (en) * | 2022-12-15 | 2023-03-28 | 福建亿山能源管理有限公司 | Operation and maintenance monitoring method and system for photovoltaic power station |
| CN116151869A (en) * | 2023-04-19 | 2023-05-23 | 国网安徽省电力有限公司经济技术研究院 | Power transmission and transformation differential operation and maintenance cost analysis system |
| CN116862482A (en) * | 2023-09-04 | 2023-10-10 | 成都昱风能源有限公司 | Power station inspection system and method based on artificial intelligence and big data analysis |
| CN116465804B (en) * | 2023-05-26 | 2023-10-20 | 东北电力大学 | System for detecting dust concentration of photovoltaic panel based on image method and application thereof |
| CN117172962A (en) * | 2023-07-17 | 2023-12-05 | 南京工业职业技术大学 | Power system-based electricity utilization energy saving method and system |
| CN117236936A (en) * | 2023-11-11 | 2023-12-15 | 中节能（新泰）太阳能科技有限公司 | Unmanned aerial vehicle-based photovoltaic power station obstacle positioning method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108038847A (en) * | 2017-12-05 | 2018-05-15 | 国网内蒙古东部电力有限公司 | Transformer inspection digital image recognition and fault detection system based on deep learning |
| CN108257114A (en) * | 2017-12-29 | 2018-07-06 | 天津市万贸科技有限公司 | A kind of transmission facility defect inspection method based on deep learning |
| CN111062991A (en) * | 2019-12-27 | 2020-04-24 | 西安因诺航空科技有限公司 | Method and device for positioning inspection target of photovoltaic scene and readable storage medium |
| CN111722642A (en) * | 2020-05-11 | 2020-09-29 | 深圳创动科技有限公司 | Inspection method and inspection device for photovoltaic power station and storage medium |
-
2021
- 2021-01-14 CN CN202110046360.5A patent/CN112633535A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108038847A (en) * | 2017-12-05 | 2018-05-15 | 国网内蒙古东部电力有限公司 | Transformer inspection digital image recognition and fault detection system based on deep learning |
| CN108257114A (en) * | 2017-12-29 | 2018-07-06 | 天津市万贸科技有限公司 | A kind of transmission facility defect inspection method based on deep learning |
| CN111062991A (en) * | 2019-12-27 | 2020-04-24 | 西安因诺航空科技有限公司 | Method and device for positioning inspection target of photovoltaic scene and readable storage medium |
| CN111722642A (en) * | 2020-05-11 | 2020-09-29 | 深圳创动科技有限公司 | Inspection method and inspection device for photovoltaic power station and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 李宁;郑仟;谢贵文;陈炜;: "基于无人机图像识别技术的输电线路缺陷检测", 电子设计工程, no. 10 * |
| 范佳楠;刘英;胡忠康;赵乾;沈鹭翔;周晓林;: "基于Faster R-CNN的实木板材缺陷检测识别系统", 林业工程学报, no. 03 * |
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113298973B (en) * | 2021-05-26 | 2022-04-26 | 广东电网有限责任公司电力调度控制中心 | Unmanned aerial vehicle power inspection method and device based on 5G communication |
| CN113298973A (en) * | 2021-05-26 | 2021-08-24 | 广东电网有限责任公司电力调度控制中心 | Unmanned aerial vehicle power inspection method and device based on 5G communication |
| CN113435369A (en) * | 2021-07-01 | 2021-09-24 | 华风数据(深圳)有限公司 | Photovoltaic fault intelligent identification method based on SSD algorithm model and far infrared video |
| CN113436176A (en) * | 2021-07-01 | 2021-09-24 | 华风数据(深圳)有限公司 | Photovoltaic panel surface fault detection method based on SSD and DSSD algorithm under visible light |
| CN113344002A (en) * | 2021-07-29 | 2021-09-03 | 北京图知天下科技有限责任公司 | Target coordinate duplication eliminating method and system, electronic equipment and readable storage medium |
| CN113325877A (en) * | 2021-08-02 | 2021-08-31 | 旻投电力发展有限公司 | RTK positioning-based photovoltaic module positioning method for unmanned aerial vehicle platform |
| CN113325878A (en) * | 2021-08-04 | 2021-08-31 | 旻投电力发展有限公司 | Photovoltaic module dust deposition detection system and detection method based on unmanned aerial vehicle |
| CN113325878B (en) * | 2021-08-04 | 2021-10-08 | 旻投电力发展有限公司 | Photovoltaic module dust deposition detection system and detection method based on unmanned aerial vehicle |
| CN114089786A (en) * | 2021-09-29 | 2022-02-25 | 北京航空航天大学杭州创新研究院 | Autonomous inspection system based on unmanned aerial vehicle vision and along mountain highway |
| CN113744270A (en) * | 2021-11-05 | 2021-12-03 | 南京市特种设备安全监督检验研究院 | Unmanned aerial vehicle visual detection and identification method for crane complex steel structure surface defects |
| CN114285374A (en) * | 2021-12-14 | 2022-04-05 | 上海电气集团股份有限公司 | Equipment positioning method and system for photovoltaic power station, electronic equipment and storage medium |
| CN114913619A (en) * | 2022-04-08 | 2022-08-16 | 华能苏州热电有限责任公司 | Intelligent mobile inspection method and system |
| CN114898232B (en) * | 2022-04-29 | 2023-08-15 | 中科云尚(南京)智能技术有限公司 | Photovoltaic power station unmanned aerial vehicle inspection method and system based on photovoltaic group string data analysis |
| CN114898232A (en) * | 2022-04-29 | 2022-08-12 | 中科云尚(南京)智能技术有限公司 | Photovoltaic power station unmanned aerial vehicle inspection method and system based on photovoltaic string data analysis |
| CN115049663A (en) * | 2022-08-16 | 2022-09-13 | 星逻智能科技(苏州)有限公司 | Unmanned aerial vehicle-based double-light fusion insulator defect detection method and system |
| CN115202404A (en) * | 2022-09-15 | 2022-10-18 | 广东容祺智能科技有限公司 | Photovoltaic power generation unit maintenance inspection method based on unmanned aerial vehicle |
| 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 |
| CN115239893A (en) * | 2022-09-23 | 2022-10-25 | 运易通科技有限公司 | Image reconstruction method for detecting defects of solar panel of warehouse ceiling |
| CN115861855A (en) * | 2022-12-15 | 2023-03-28 | 福建亿山能源管理有限公司 | Operation and maintenance monitoring method and system for photovoltaic power station |
| CN115861855B (en) * | 2022-12-15 | 2023-10-24 | 福建亿山能源管理有限公司 | Operation and maintenance monitoring method and system for photovoltaic power station |
| CN115767040A (en) * | 2023-01-06 | 2023-03-07 | 松立控股集团股份有限公司 | 360-degree panoramic monitoring automatic cruise method based on interactive continuous learning |
| CN116151869B (en) * | 2023-04-19 | 2023-06-27 | 国网安徽省电力有限公司经济技术研究院 | Power transmission and transformation differential operation and maintenance cost analysis system |
| CN116151869A (en) * | 2023-04-19 | 2023-05-23 | 国网安徽省电力有限公司经济技术研究院 | Power transmission and transformation differential operation and maintenance cost analysis system |
| CN116465804B (en) * | 2023-05-26 | 2023-10-20 | 东北电力大学 | System for detecting dust concentration of photovoltaic panel based on image method and application thereof |
| CN117172962A (en) * | 2023-07-17 | 2023-12-05 | 南京工业职业技术大学 | Power system-based electricity utilization energy saving method and system |
| CN117172962B (en) * | 2023-07-17 | 2024-04-16 | 南京工业职业技术大学 | Power system-based electricity utilization energy saving method and system |
| CN116862482A (en) * | 2023-09-04 | 2023-10-10 | 成都昱风能源有限公司 | Power station inspection system and method based on artificial intelligence and big data analysis |
| CN116862482B (en) * | 2023-09-04 | 2023-11-07 | 成都昱风能源有限公司 | Power station inspection system and method based on artificial intelligence and big data analysis |
| CN117236936A (en) * | 2023-11-11 | 2023-12-15 | 中节能（新泰）太阳能科技有限公司 | Unmanned aerial vehicle-based photovoltaic power station obstacle positioning method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112633535A (en) | Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle image | |
| CN106203265B (en) | A kind of Construction Fugitive Dust Pollution source monitors automatically and coverage forecasting system and method | |
| CN109977813B (en) | Inspection robot target positioning method based on deep learning framework | |
| CN103941746B (en) | Image processing system and method is patrolled and examined without man-machine | |
| CN108734143A (en) | A kind of transmission line of electricity online test method based on binocular vision of crusing robot | |
| CN112327906A (en) | Intelligent automatic inspection system based on unmanned aerial vehicle | |
| CN106815838A (en) | A kind of method and system of the detection of photovoltaic module hot spot | |
| CN111582234B (en) | Large-scale oil tea tree forest fruit intelligent detection and counting method based on UAV and deep learning | |
| CN113781450A (en) | Automatic intelligent defect analysis system based on unmanned aerial vehicle image acquisition of power transmission and distribution line | |
| CN113075938A (en) | Remote intelligent inspection system and method for power transmission line | |
| CN111241615A (en) | Highly realistic multi-source fusion three-dimensional modeling method for transformer substation | |
| CN115240093B (en) | Automatic power transmission channel inspection method based on visible light and laser radar point cloud fusion | |
| CN114240868A (en) | Unmanned aerial vehicle-based inspection analysis system and method | |
| CN113743358B (en) | Landscape vision feature recognition method adopting omnibearing collection and intelligent calculation | |
| CN116258980A (en) | Unmanned aerial vehicle distributed photovoltaic power station inspection method based on vision | |
| CN113378754B (en) | Bare soil monitoring method for construction site | |
| CN113033386B (en) | High-resolution remote sensing image-based transmission line channel hidden danger identification method and system | |
| CN113111714A (en) | Risk prevention and control system applied to power transmission field operation | |
| CN212515006U (en) | Accurate positioner of unusual target suitable for unmanned aerial vehicle photovoltaic patrols and examines scene | |
| CN112184903A (en) | Method, device, equipment and medium for detecting high-voltage line tree obstacle risk points | |
| Sha et al. | Design of patrol monitoring and control system for hot spot of solar photovoltaic module | |
| CN116070523A (en) | Simulation construction method for substation defect detection image sample | |
| CN114549432A (en) | Intelligent unmanned aerial vehicle image identification method | |
| CN110658844B (en) | Ultra-high voltage direct current line channel unmanned aerial vehicle monitoring method and system | |
| CN114285374A (en) | Equipment positioning method and system for photovoltaic power station, electronic equipment and storage medium |
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 |