CN110378903A - A kind of transmission line of electricity anti-accident measures Intelligent statistical method - Google Patents
A kind of transmission line of electricity anti-accident measures Intelligent statistical method Download PDFInfo
- Publication number
- CN110378903A CN110378903A CN201910869429.7A CN201910869429A CN110378903A CN 110378903 A CN110378903 A CN 110378903A CN 201910869429 A CN201910869429 A CN 201910869429A CN 110378903 A CN110378903 A CN 110378903A
- Authority
- CN
- China
- Prior art keywords
- picture
- hanging point
- type
- training
- steel tower
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of transmission line of electricity anti-accident measures Intelligent statistical methods;Include: S1: input needs the electric power line pole tower photo identified, S2: traversal electric power line pole tower picture;S3: the steel tower type and hanging point type of picture are identified;S4: identification steel tower type and hanging point type are obtained if identifying successfully, and is marked in interception picture, picture is saved, while recording recognition result, into S6;Enter S5 if recognition failures;S5: judging whether picture traverses terminates;If no, returning to S2;If completing, and equal recognition failures, then saves the original image of nonrecognition and recording recognition result, into S6;S6: end of identification exports the component picture and nonrecognition picture of recognition result, interception.S7: S8: secondary identification according to scissors crossing information and image recognition result, counts anti-situation of arranging.Transmission line of electricity anti-accident measures statistics of the present invention is comprehensive, high-efficient.
Description
Technical field
The present invention relates to high voltage transmission line road transports to examine field, more particularly, to a kind of transmission line of electricity anti-accident measures intelligence
It can statistical method.
Background technique
In recent years, as economic development power supply demand also increases year by year, ultra-high-tension power transmission line is steady as guarantee power grid
Surely the main artery run, significance are self-evident.In order to guarantee the safe and stable operation of ultra-high-tension power transmission line, transmission line of electricity
O&M unit has formulated a series of stringent anti-accident measures, and all transmission lines of electricity for meeting correlated condition all must be according to formulation
Measure is rectified and improved, to improve the general safety coefficient of transmission line of electricity.However existing transmission line of electricity anti-accident measures implementation is deposited
At following three aspect the problem of: one, it is not comprehensive to need to carry out the anti-route statistical information arranged, and there are mistakes and omissions, causing to exist should
Implement anti-accident measures and practical the case where rectifying and improving without development.Transmission line of electricity is distributed multi-point and wide-ranging, traditional power transmission line equipment
Information statistical method can only by artificial observation, take pictures or inspection information statistic record, obtain and need by manually counting to summarize
Implement the power transmission line facility information of anti-accident measures, however artificial statistics tends to mistakes and omissions occur.Through investigating, an O&M
Personnel generally require to undertake the O&M responsibility of four or five ten kilometers of transmission lines of electricity, and overweight work load adds operation maintenance personnel quality
Irregular, groups of people have little understanding to the implementation condition of anti-accident measures, therefore that it is easy to appear statistical data is not comprehensive, deposit
It is practical without carrying out the case where rectifying and improving in the implementation anti-accident measures, cause the security risk of equipment.Two, counter that work is arranged to imitate
Rate is low, takes long time, and cannot handle security risk in time, guarantees the safe and stable operation of route.It is traditional anti-information to be arranged to unite
Meter needs to carry out by artificial, and statistics, which summarizes, to be obtained, therefore generally requires to be responsible for personnel and route operation maintenance personnel ditch repeatedly specially
It is logical to verify statistical information, also need to modify repeatedly there are mistakes and omissions situation, this resulted in it is counter arrange work to take long time, inefficiency is asked
Topic, cannot arrange the rectification of route hidden danger in time, guarantee the anti-timeliness for arranging implementation.Three, there is repeat statistical information, waste
The phenomenon that human and material resources.Anti- facility information is arranged manually to count the problem of being easy to appear mistakes and omissions, route O&M due to traditional
Unit carries out multiple line facility information to guarantee that the correct of statistical information reliably often requires that operation maintenance personnel repeats to count
Checking work, therefore it is low to cause anti-accident measures working efficiency, needs to carry out identical statistical work repeatedly, both occupy
The time of operation maintenance personnel causes unnecessary work load, also wastes a lot of manpower and material resources resource.Therefore a kind of spirit is needed
Efficient transmission line of electricity anti-accident measures Intelligent statistical method living saves a large amount of manpower object to improve the anti-working efficiency arranged
While power resource, guarantees the anti-comprehensive development arranged, improve the general safety coefficient of transmission line of electricity, guarantee the safety and stability of route
Operation.
Summary of the invention
The present invention is to overcome the statistics of transmission line of electricity anti-accident measures described in the above-mentioned prior art not comprehensive, low efficiency scarce
It falls into, a kind of transmission line of electricity anti-accident measures Intelligent statistical method is provided.
It the described method comprises the following steps:
S1: input needs the electric power line pole tower photo identified,
S2: using the detection window of fixed size, traversing electric power line pole tower picture, and the picture for intercepting fixed size carries out
Identification;
S3: identifying the picture of S2 interception, the Fast R-CNN steel tower type and hanging point identification model of foundation training early period,
The steel tower type and hanging point type of picture are identified;
S4: judge recognition result;Identification steel tower type and hanging point type are obtained if identifying successfully, and are carried out in interception picture
Label saves picture, while recording recognition result, into S6;Enter S5 if recognition failures;
S5: judging whether picture traverses terminates;If not traversing completion, returns to the other pictures of S2 interception and identified;If
Picture traverses completion, and equal recognition failures, saves the original image of nonrecognition and is recording recognition result, into S6;
S6: end of identification exports the component picture and nonrecognition picture of recognition result, interception.It can people for the result of nonrecognition
Work differentiates and mark, improves data set, incremental training more new model.For recognition result, again by the component picture of interception
Input model is recognized, into S7.
S7: secondary identification, to S4 interception picture be recognized, according to early period training yolov3 steel tower type and
Hanging point type identification model is identified, the component of identification is marked with rectangle frame, and record recognition result, is entered
S8;For the result of nonrecognition can artificial cognition and mark, improve data set, incremental training more new model;
S8: according to scissors crossing information and image recognition result, anti-situation of arranging is counted.
Preferably, the electric power line pole tower photo for needing to identify in S1 includes the ground wire hanging point of daily O&M unmanned plane shooting
Photo and the ground wire hanging point photo manually shot.
Preferably, electric power line pole tower picture is traversed in S2 specifically: from the electric power line pole tower photo for needing to identify
Between, upper left side, five upper right side, lower left and lower right directions traverse electric power line pole tower picture.
Preferably, it is identified for the first time in S2 from the middle of original image and intercepts picture recognition, work as recognition failures, then successively cut
The picture in other directions is taken to be identified.
Preferably, the training process of Fast R-CNN steel tower type and hanging point identification model the following steps are included:
S3.1: mark picture, formed data set: establish steel tower type and ground wire hanging point categorical data collection, to daily O&M nobody
The ground wire hanging point photo of machine shooting and the ground wire hanging point photo manually shot are labeled, and are hung using labelImg software to ground wire
Vertex type is labeled, and is respectively labeled as tangent tower, anchor support, single hanging point and double hanging point, training dataset is consequently formed;
S3.2: model training forms steel tower type and hanging point identification model: using PaddlePaddle deep learning platform and
Fast R-CNN algorithm of target detection obtains steel tower type and hanging point identification model by 60,000 times or more repetitive exercises;
S3.3: input electric power line pole tower photo array judgement: the steel tower type and hanging point identification model obtained according to training is right
The picture of input is identified;
S3.4: judging whether there is nonrecognition as a result, illustrating that training pattern is closed if all pictures of input identify success
Lattice, discrimination are higher;Nonrecognition improves data set, returns to S3.2, increment instruction as a result, then need artificial cognition and mark if it exists
Practice and form new model, recycling input picture re-recognizes, and thus constantly improve training pattern, obtains accurate picture training mould
Type.
The features of the present invention is mainly reflected in first is that how to accurately identify to the tiny component of transmission line of electricity.The present invention mentions
Traversal picture interception technology is gone out, big figure is resolved into small figure and is identified, improved discrimination, specific steps are shown in S2.Second is that mentioning
Fast R-CNN model identification component is first used in secondary identification out, intercepts the component part identified in picture, then the figure to interception
Piece is recognized, and is recognized with yolov3 model, is screened out wrong as a result, into one in wrong identification for the first time
Step improves the accuracy rate of identification.Third is that the application to recognition result, transmission line of electricity distribution is multi-point and wide-ranging, and an area is up to ten thousand easily
How base shaft tower effectively applies so many shaft tower recognition result, the invention proposes the data to critical section shaft tower,
If the shaft tower recognition result of important scissors crossing, populated area extracts, the key line of operation maintenance personnel concern is generated
Section recognition result has used the algorithm compared, and program is automatically according to ready route inventory, from magnanimity recognition result data
In select result data in inventory, form final result, specific steps are shown in S8, to guarantee the utilization rate and work of recognition result
Crucial point is played with real.
Compared with prior art, the beneficial effect of technical solution of the present invention is:
(1) counter to arrange Information Statistics more objective comprehensively.The invention is based on more mature image recognition technology, by power transmission line
The intelligent recognition of line pole tower photo, can automatic discrimination whether meet and implement anti-accident measures condition, sentence with traditional artificial count
Disconnected comprehensive compared to more objective, error rate and missing rate are lower, more can guarantee that counter arrange is implemented as required.
(2) shortening is counter arranges the working time, improves anti-working efficiency of arranging.The invention provides a kind of anti-accidents of transmission line of electricity to arrange
Intelligent statistical method is applied, system programming count obtains transmission line of electricity anti-accident measures performance, eliminates the phase manually counted
It mutually links up, exchanges verification process, the plenty of time can be saved, guarantee the timely implementation of transmission line of electricity anti-accident measures, improve work
Make efficiency.
(3) system programming count, can constantly upgrade perfect, can save a large amount of human and material resources.What the invention provided
Intelligent statistical method, system programming count are obtained a result, and are only needed artificial cognition identification error message and are marked, can sophisticated systems
Function, the accuracy of constantly improve statistical information.Compared with traditional artificial verification of statistics repeatedly, reduces and manually count repeatedly
The workload of verification saves a large amount of human and material resources.
The present invention effectively increases the intelligence, automation level of transmission line of electricity maintenance work, and there is popularization well to answer
With value.
Detailed description of the invention
Fig. 1 is a kind of transmission line of electricity anti-accident measures Intelligent statistical method flow diagram.
Fig. 2 is image recognition model training flow chart.
Specific embodiment
The attached figures are only used for illustrative purposes and cannot be understood as limitating the patent;
In order to better illustrate this embodiment, the certain components of attached drawing have omission, zoom in or out, and do not represent the ruler of actual product
It is very little;
To those skilled in the art, the omitting of some known structures and their instructions in the attached drawings are understandable.
The following further describes the technical solution of the present invention with reference to the accompanying drawings and examples.
The present embodiment provides a kind of transmission line of electricity anti-accident measures Intelligent statistical methods.In this example, it is assumed that needing
The important scissors crossing steel tower type of transmission line of electricity and ground wire hanging point type are counted, if route is in important scissors crossing section
Inside there is the case where straight line pole ground wire list hanging point, then needs as required into development transmission line of electricity anti-accident measures.
As shown in Figure 1, the described method comprises the following steps:
S1: input shaft tower photo.Input needs the electric power line pole tower photo that identifies, input in the present embodiment daily O&M without
The ground wire hanging point photo of man-machine shooting and the ground wire hanging point photo manually shot are differentiated, and are recorded according to photo name classification
Recognition result.
S2: fixed test window size, traversal picture interception are identified.To guarantee image recognition effect, need to intercept
The picture of fixed size is identified.In the present embodiment with the detection window of 2000 pixel *, 2000 pixel size from original
Among beginning picture, upper left side, upper right side, five directions such as lower left and lower right traverse electric power line pole tower pictures, interception
The image block of 2000 pixel *, 2000 pixel size is identified.To improve recognition efficiency, first identification by from original image just
Taken intermediate picture recognition works as recognition failures, then successively intercepts the picture in other directions and identified.
S3: steel tower type and hanging point identification.In the present embodiment, the picture of S2 interception is identified, is instructed according to early period
Experienced Fast R-CNN steel tower type and hanging point type identification model, identifies the steel tower type and hanging point type of picture.
S4: judge recognition result.The steel tower type and hanging point type of identification are obtained if identifying successfully, and are intercepted in picture
Identified component, is marked with rectangle frame, picture is saved, while recognition result is recorded in Microsoft Excel, into S6.
Enter S5 if recognition failures.
Component in label digit synbol picture described in the present embodiment, as by the component identified in picture with the shape of rectangle frame
Formula is marked.
Image recognition in the present embodiment refers to one kind in image recognition, object detection, and the result of object detection is usual
With the formal notation of rectangle frame object detected.
S5: judging whether picture traverses terminates.If not traversing completion, the picture progress that S2 intercepts other directions is returned
Identification.If the picture in five directions traverses completion, and equal recognition failures, the original image of nonrecognition is saved, and in EXCEL
Recognition result is recorded in table, into S6.
Wherein table 1 is iron tower of power transmission line type and hanging point types of image recognition result summary sheet.
1 iron tower of power transmission line type of table and hanging point types of image recognition result summary sheet
S6: end of identification exports the component picture and nonrecognition picture of recognition result, interception.It can people for the result of nonrecognition
Work differentiates and mark, improves data set, incremental training, more new model.For recognition result, again by the component picture of interception
Input model is recognized, into S7.
More new model described in the present embodiment i.e. incremental training on the basis of old model, redefines a series of ginsengs of model
Number, improves old model to reach, promotes the effect of discrimination.
S7: secondary identification, to S4 interception picture be recognized, according to early period training yolov3 steel tower type and
Hanging point type identification model is identified, the component of identification is marked with rectangle frame, and is recorded and known in Microsoft Excel
Not as a result, the recognition result for being aggregated to form electric power line pole tower photo summarizes table, into S8.It can people for the result of nonrecognition
Work differentiates and mark, improves data set, incremental training more new model.
Fast R-CNN steel tower type described in the present embodiment and hanging point type identification model refer to Fast R-CNN target
The convolutional neural networks model that detection algorithm training generates, yolov3 steel tower type and hanging point type identification model refer to utilization
The convolutional neural networks model that the training of yolov3 algorithm of target detection generates.Yolov3 model and Fast R-CNN model are objects
Body detection model, the algorithm only used is different, but yolov3 recognition speed is faster, comprehensively considers and carries out using yolov3
Secondary identification can improve recognition efficiency on the basis of guaranteeing recognition speed.
S8: according to scissors crossing information and image recognition result, anti-situation of arranging is counted.In the present embodiment, it needs to defeated
The important scissors crossing steel tower type of electric line and ground wire hanging point type are counted, the transmission line of electricity weight counted according to daily O&M
Scissors crossing section statistical table is wanted, the line steel tower type and hanging point type statistics table that as shown in table 2 and S6 summarizes, as above
Shown in table 1, system is compared automatically.Specific comparison method is, important scissors crossing section steel tower wherein one in first consult table 2
The information of one base shaft tower of route, then the tower of corresponding line shaft tower and hanging point type information in consult table 1, by comparing, if
Overhead line structures the case where there are straight line pole ground wire list hanging points, system automatically extracts to form the overhead line structures section and shaft tower again
Relevant information, be recorded in Microsoft Excel, system by circulation consults comparison, and can to form fruit anti-accident measures to be performed clear
It is single, as shown in table 3, transmission line of electricity anti-accident measures Intelligent statistical is completed with this.
The important scissors crossing section inventory of 2 transmission line of electricity of table
The anti-accident measures inventory to be performed of table 3
As shown in Fig. 2, Fig. 2 is image recognition model training flow chart described in the present embodiment.In this example, it is assumed that needing
Iron tower of power transmission line type and ground wire hanging point type identification model are trained.
What model training described in the present embodiment was used is convolutional neural networks model, and the training of model has series of steps,
Including convolution, nonlinear transformation, pond or sub-sampling and classification etc., the training of model is exactly first to define nerve net in simple terms
The structure of network model, then the various parameters of the training Definition Model by mass data collection finally train complete convolution mind
Through network model.A series of aforementioned described processing can be carried out to the picture of input with the model, and picture is searched by processing
In whether have component predetermined, identify.
Specific Fast R-CNN model and yolov3 model training workflow the following steps are included:
S3.1: mark picture forms data set.Need to establish steel tower type and ground wire hanging point categorical data in the present embodiment
Collection, the ground wire hanging point photo shoot to daily O&M unmanned plane and the ground wire hanging point photo manually shot are labeled, and are used
LabelImg software is labeled ground wire hanging point type, as shown in table 4, be respectively labeled as tangent tower, anchor support, single hanging point and
Training dataset is consequently formed in double hanging point.
Table 4: the steel tower type and ground wire hanging point type classification for needing to mark
S3.2: model training forms steel tower type and hanging point identification model.Need in the present embodiment train steel tower type and
Ground wire hanging point type identification model, the model that primary identification is generated using the training of Fast R-CNN algorithm of target detection, tool
PaddlePaddle deep learning platform and Fast R-CNN algorithm of target detection are utilized for body, pass through convolution, non-linear change
It changes, pond or sub-sampling and classification etc. series of steps training convolutional neural networks model, eventually by 60,000 times or more
Repetitive exercise obtains Fast R-CNN steel tower type and hanging point identification model.What it is for secondary identification is yolov3 target detection
The model that algorithm training generates, is specifically instructed using PaddlePaddle deep learning platform and yolov3 algorithm of target detection
Practice and generates convolutional neural networks model.
S3.3: input electric power line pole tower photo array judgement.Steel tower type and hanging point the identification mould obtained according to training
Type identifies the picture of input.
S3.4: it whether there is nonrecognition result.If input all pictures identify success, illustrate training pattern compared with
Good, discrimination is higher;Nonrecognition improves data set, returns to S2, incremental training shape as a result, need artificial cognition and mark if it exists
At new model, recycling input picture is re-recognized, and is thus constantly improve training pattern, is obtained accurate picture training mould
Type.
The terms describing the positional relationship in the drawings are only for illustration, should not be understood as the limitation to this patent;
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be to this hair
The restriction of bright embodiment.For those of ordinary skill in the art, it can also do on the basis of the above description
Other various forms of variations or variation out.There is no necessity and possibility to exhaust all the enbodiments.It is all in the present invention
Spirit and principle within made any modifications, equivalent replacements, and improvements etc., should be included in the guarantor of the claims in the present invention
Within the scope of shield.
Claims (5)
1. a kind of transmission line of electricity anti-accident measures Intelligent statistical method, is characterized in that, the described method comprises the following steps:
S1: input needs the electric power line pole tower photo identified,
S2: using the detection window of fixed size, traversing electric power line pole tower picture, and the picture for intercepting fixed size carries out
Identification;
S3: identifying the picture of S2 interception, the Fast R-CNN steel tower type and hanging point identification model of foundation training early period,
The steel tower type and hanging point type of picture are identified;
S4: judge recognition result;Identification steel tower type and hanging point type are obtained if identifying successfully, and are carried out in interception picture
Label saves picture, while recording recognition result, into S6;Enter S5 if recognition failures;
S5: judging whether picture traverses terminates;If not traversing completion, returns to the other pictures of S2 interception and identified;If
Picture traverses completion, and equal recognition failures, saves the original image of nonrecognition and is recording recognition result, into S6;
S6: end of identification exports the component picture and nonrecognition picture of recognition result, interception;The result of nonrecognition is passed through
Artificial cognition and mark, improve data set, incremental training more new model;For recognition result, again by the component picture of interception
Secondary input model, is recognized, into S7;
S7: secondary identification is recognized the picture of S4 interception, the yolov3 steel tower type and hanging point according to training early period
Type identification model is identified, the component of identification is marked with rectangle frame, and record recognition result, into S8;It is right
It by artificial cognition and is marked in the result of nonrecognition, improves data set, incremental training more new model;
S8: according to scissors crossing information and image recognition result, anti-situation of arranging is counted.
2. transmission line of electricity anti-accident measures Intelligent statistical method according to claim 1, is characterized in that, needs to identify in S1
Electric power line pole tower photo include the ground wire hanging point photo of daily O&M unmanned plane shooting and the ground wire hanging point photograph manually shot
Piece.
3. transmission line of electricity anti-accident measures Intelligent statistical method according to claim 2, is characterized in that, transmission of electricity is traversed in S2
Overhead line structures picture specifically: from need among the electric power line pole tower photo that identifies, upper left side, upper right side, lower left and the right side
The direction of lower section five traverses electric power line pole tower picture.
4. transmission line of electricity anti-accident measures Intelligent statistical method according to claim 3, is characterized in that, identifies for the first time in S2
Picture recognition is intercepted from the middle of original image, works as recognition failures, then successively intercepts the picture in other directions and is identified.
5. transmission line of electricity anti-accident measures Intelligent statistical method according to claim 1-4, is characterized in that, Fast
The training process of R-CNN steel tower type and hanging point identification model the following steps are included:
S3.1: mark picture, formed data set: establish steel tower type and ground wire hanging point categorical data collection, to daily O&M nobody
The ground wire hanging point photo of machine shooting and the ground wire hanging point photo manually shot are labeled, using labelImg software to steel tower class
Type and ground wire hanging point type are labeled, and are respectively labeled as tangent tower, anchor support, single hanging point and double hanging point, training is consequently formed
Data set;
S3.2: model training forms steel tower type and hanging point identification model: using PaddlePaddle deep learning platform and
Fast R-CNN algorithm of target detection obtains steel tower type and hanging point identification model by 60,000 times or more repetitive exercises;
S3.3: input electric power line pole tower photo array judgement: the steel tower type and hanging point identification model obtained according to training is right
The picture of input is identified;
S3.4: judging whether there is nonrecognition as a result, illustrating that training pattern is closed if all pictures of input identify success
Lattice, discrimination are higher;Nonrecognition improves data set, returns to S3.2, increment instruction as a result, then need artificial cognition and mark if it exists
Practice and form new model, recycling input picture re-recognizes, and thus constantly improve training pattern, obtains accurate picture training mould
Type.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910869429.7A CN110378903A (en) | 2019-09-16 | 2019-09-16 | A kind of transmission line of electricity anti-accident measures Intelligent statistical method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910869429.7A CN110378903A (en) | 2019-09-16 | 2019-09-16 | A kind of transmission line of electricity anti-accident measures Intelligent statistical method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110378903A true CN110378903A (en) | 2019-10-25 |
Family
ID=68261538
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910869429.7A Pending CN110378903A (en) | 2019-09-16 | 2019-09-16 | A kind of transmission line of electricity anti-accident measures Intelligent statistical method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110378903A (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108389197A (en) * | 2018-02-26 | 2018-08-10 | 上海赛特斯信息科技股份有限公司 | Transmission line of electricity defect inspection method based on deep learning |
CN109389161A (en) * | 2018-09-28 | 2019-02-26 | 广州大学 | Rubbish identification evolutionary learning method, apparatus, system and medium based on deep learning |
CN109544533A (en) * | 2018-11-23 | 2019-03-29 | 聚时科技(上海)有限公司 | A kind of metal plate defect detection and measure based on deep learning |
CN109785361A (en) * | 2018-12-22 | 2019-05-21 | 国网内蒙古东部电力有限公司 | Substation's foreign body intrusion detection system based on CNN and MOG |
CN110119686A (en) * | 2019-04-17 | 2019-08-13 | 电子科技大学 | A kind of safety cap real-time detection method based on convolutional neural networks |
-
2019
- 2019-09-16 CN CN201910869429.7A patent/CN110378903A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108389197A (en) * | 2018-02-26 | 2018-08-10 | 上海赛特斯信息科技股份有限公司 | Transmission line of electricity defect inspection method based on deep learning |
CN109389161A (en) * | 2018-09-28 | 2019-02-26 | 广州大学 | Rubbish identification evolutionary learning method, apparatus, system and medium based on deep learning |
CN109544533A (en) * | 2018-11-23 | 2019-03-29 | 聚时科技(上海)有限公司 | A kind of metal plate defect detection and measure based on deep learning |
CN109785361A (en) * | 2018-12-22 | 2019-05-21 | 国网内蒙古东部电力有限公司 | Substation's foreign body intrusion detection system based on CNN and MOG |
CN110119686A (en) * | 2019-04-17 | 2019-08-13 | 电子科技大学 | A kind of safety cap real-time detection method based on convolutional neural networks |
Non-Patent Citations (1)
Title |
---|
陈慧元 等: "基于级联卷积神经网络的大场景遥感图像舰船目标快速检测方法", 《雷达学报》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20170308934A1 (en) | Management method of power engineering cost | |
CN109146419A (en) | Drawing classification auditing system based on artificial intelligence | |
CN107862450A (en) | Performance appraisal system and method | |
CN104077657B (en) | A kind of Information Evaluation method based on quantizating index | |
CN111754020A (en) | Intelligent power plant equipment inspection system and method based on data analysis | |
CN109829497A (en) | A kind of area user identification zone and method of discrimination based on supervised learning | |
CN110309884A (en) | Electricity consumption data anomalous identification system based on ubiquitous electric power Internet of Things net system | |
CN114155072B (en) | Financial prediction model construction method and system based on big data analysis | |
CN105574593A (en) | Track state static-state detection and control system and method based on cloud computing and big data | |
CN105574666A (en) | Method and device for evaluating credit level of enterprise based on key data modeling | |
CN110991985A (en) | Project progress data processing method | |
CN110705887A (en) | Low-voltage transformer area operation state comprehensive evaluation method based on neural network model | |
CN110059913A (en) | A kind of quantitative estimation method counted and the power failure of future-state is planned | |
CN110968703B (en) | Method and system for constructing abnormal metering point knowledge base based on LSTM end-to-end extraction algorithm | |
CN115034671A (en) | Secondary system information fault analysis method based on association rule and cluster | |
CN110309737A (en) | A kind of information processing method applied to cigarette sales counter, apparatus and system | |
CN110490220A (en) | A kind of bus load discrimination method and system | |
CN110378903A (en) | A kind of transmission line of electricity anti-accident measures Intelligent statistical method | |
CN117520184A (en) | Test system for developing computer software | |
CN110261693A (en) | A kind of online thermal test method and system based on stable state sampling technique | |
CN115719139A (en) | Dispatching self-checking system for power grid dispatching operation management | |
CN112732773B (en) | Method and system for checking uniqueness of relay protection defect data | |
CN114757472A (en) | Comprehensive scoring system for personnel management | |
CN114911847B (en) | Cross-border trade data statistics and analysis method based on big data | |
CN207516464U (en) | A kind of Portable metering automatization terminal trouble-shooter |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191025 |
|
RJ01 | Rejection of invention patent application after publication |