CN108564565A - A kind of power equipment infrared image multi-target orientation method based on deep learning - Google Patents
A kind of power equipment infrared image multi-target orientation method based on deep learning Download PDFInfo
- Publication number
- CN108564565A CN108564565A CN201810199434.7A CN201810199434A CN108564565A CN 108564565 A CN108564565 A CN 108564565A CN 201810199434 A CN201810199434 A CN 201810199434A CN 108564565 A CN108564565 A CN 108564565A
- Authority
- CN
- China
- Prior art keywords
- infrared image
- power equipment
- deep learning
- training
- method based
- 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
- 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
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06N3/047—Probabilistic or stochastic 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared 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
Abstract
The invention discloses a kind of power equipment infrared image multi-target orientation method based on deep learning, including step：1) standardized power equipment infrared image is obtained by substation equipment detection device；2) power equipment infrared image sample database, extraction training set, verification collection and test set are established；3) FASTER RCNN depth targets detection neural network is established, the FASTER RCNN depth targets detection neural network established is trained using the training set of sample database, and the over-fitting degree by verifying the set pair analysis model is verified；4) network model established using training is carried out multi-targets recognition and positioning to the infrared image in test set, and generates recognition result.The present invention carries out depth characteristic excavation using deep learning algorithm to input infrared image, independent of manual extraction characteristic parameter, the region and position of all kinds of electrical power mains equipment can effectively and be accurately identified and position, to a certain extent, reduce manual labor amount.
Description
Technical field
The present invention relates to the technical fields of power equipment infrared image identification and positioning, refer in particular to a kind of based on depth
The power equipment infrared image multi-target orientation method of habit.
Background technology
Power equipment infrared image is the infrared energy sent out by infrared technique by detecting power equipment, and is converted
For corresponding electric signal, power equipment surface thermal image is obtained after Electric signal processing.Infrared detection technology have it is remote,
Do not contact, do not sample, not disintegrating, accurately, quickly, it is intuitive the features such as, be widely used in power equipment overheating defect detect and diagnose
In, the stability to improving electric system is of great significance.Therefore, it effectively and accurately identifies in power equipment infrared image
Electrical power mains equipment region and position become one ring of key applied in the power system for infrared technique.
In infrared image fault diagnosis, it is desirable to be able to which obtaining multiple target areas in infrared image, (for example three-phase is set
Position where standby or same category of device), in each extracted region maximum temperature information, the shape of power equipment is compared by temperature
State.Image segmentation and images match are two kinds of common methods in target positioning, and wherein Da-Jin algorithm and watershed method is as image
The representative of split plot design, by reaching target identification effect into row threshold division to power equipment infrared image.Images match is meter
The research hotspot of calculation machine vision, image procossing and area of pattern recognition, method are broadly divided into two classes：Matching process based on region
With feature-based matching method.But infrared image is a kind of pseudo- color image, it reflects the height of body surface temperature
And distribution, there are the characteristics such as low with contrast in strength set.And it is limited by infrared thermoviewer technology, the quality of infrared image
It is usually not high.Therefore image partition method tends not to target and background to split, and is even more difficulty for Segmentation of Multi-target.
In addition, infrared picture data amount is big and form is changeable, images match method is easy to be limited to single matching template so as to cause logical
It is not strong with property.In order to realize power equipment infrared image multi-targets recognition and positioning, the present invention innovate by deep learning application
Into image procossing, it is obviously improved recognition efficiency and accuracy.
Deep study is used as one branch of machine learning, it is intended to the structure of structure deeper to reinforce model in mass data
The middle ability for capturing hidden feature.With traditional shallow-layer neural network comparatively, deep learning have it is a series of can carry out it is non-
The hidden layer of linear transformation, so as to challenge more complex environment and problem.Infrared image target identification is asked with positioning
For topic, the convolutional neural networks in deep learning can obtain the feature statement of image deeper.Therefore, deep learning is managed
It is the important support of electricity power Transformation Development and the important trend of electric power network technique development by field of power is applied to.
Invention content
The shortcomings that it is an object of the invention to overcome the prior art and deficiency, it is proposed that a kind of electric power based on deep learning
Equipment infrared image multi-target orientation method, based in deep learning FASTER-RCNN depth targets detection neural network come
The deep-neural-network model of end-to-end training is built, can more accurately and efficiently be identified in power equipment infrared image
The region and position of electrical power mains equipment.
To achieve the above object, technical solution provided by the present invention is：A kind of power equipment based on deep learning is red
Outer image multi-target orientation method, includes the following steps：
Step 1：Standardized power equipment infrared image is obtained by substation equipment detection device；
Step 2：Establish power equipment infrared image sample database, extraction training set, verification collection and test set；
Step 3：FASTER-RCNN depth targets detection neural network is established, using the training set of sample database to being established
FASTER-RCNN depth targets detection neural network be trained, and pass through verify the set pair analysis model over-fitting degree carry out
Verification constantly obtains the connection weight and offset parameter of network model after debugging；
Step 4：The network model established using training carries out multi-targets recognition, to red to the infrared image in test set
The region and position of trunk power equipment extract in outer collection of illustrative plates, and generate the recognition result of power equipment infrared image.
In step 1, pass through substation technical staff hand-held thermal imager collection in worksite picture or Intelligent Mobile Robot
Thermal infrared imager is carried in inspection track photographs infrared image, wherein the power equipment for including in the infrared image acquired
Have：Power transformer, breaker, arrester, voltage sensor, current sensor.
In step 2, collected power equipment infrared image is divided into training set according to preset ratio, verification collects and surveys
Examination collection, and the infrared image concentrated to training set and verification adds label, is fabricated to the data set of PASCAL VOC formats,
In, the sample image of test set requires cannot be Chong Die with the image that verification is concentrated with training set, and need not add label.
In step 3, the FASTER-RCNN depth targets established detect neural network：For extracting image spy
The convolutional neural networks (Convolutional Neural Network, CNN) of characteristic spectrum are levied and generate, for characteristic pattern
Each pixel in spectrum generates the region candidate network (Region Proposal Network, RPN) of candidate region, with
And the interest pool area layer (Region Of Interests, ROI) for being combined characteristics of image with selected candidate region；
Optimal region is selected finally by the frame Return Law (Bounding-Box, Bbox), and then completes target recognition and classification.
In step 4, the FASTER-RCNN depth targets infrared image input step 3 in test set built are examined
Neural network is surveyed, the region and position of electrical power mains equipment in infrared image are extracted and obtain power equipment infrared image
Recognition result.
The convolutional neural networks structure for forming FASTER-RCNN carries out pre-training by Imagenet training sets, corresponding
Convolutional layer activation primitive is linear amending unit function (Rectified Linear Unit, ReLU).
The gradient optimal method that the FASTER-RCNN depth targets detection neural network uses is calculated for adaptability moments estimation
Method (Adaptive Moment Estimation, Adam), in an iterative process, the variation of observation loss function value judge convergence
Situation, regularized learning algorithm rate, and the over-fitting in training process is reduced using dropout methods in the network.
Compared with prior art, the present invention having the following advantages that and advantageous effect：
By the present invention in that depth characteristic excavation is carried out to input infrared image with convolutional neural networks, independent of craft
Feature extraction parameter, and the over-fitting in training process is reduced by being introduced into dropout technologies, so as to effectively simultaneously
The region and position of all kinds of electrical power mains equipment are accurately identified and positioned, and then is reduced to a certain extent to manual identified
Dependence, reduce manual labor amount.
Description of the drawings
Fig. 1 is the overview flow chart of the method for the present invention.
Fig. 2 is the basic structure schematic diagram of a two-dimensional convolution neural network.
Fig. 3 is the structure chart that FASTER-RCNN depth targets detect neural network.
Specific implementation mode
The present invention is further explained in the light of specific embodiments.
The present invention is for the positioning of current power equipment infrared image and recognition accuracy is not high, multi-targets recognition is difficult, people
A kind of the problems such as work degree of dependence is high, it is proposed that power equipment infrared image multi-target orientation method based on deep learning.
In embodiment, using power equipment infrared image as input, by deep learning method to the region and position of electrical power mains equipment
It sets and is positioned and identified, the scheme of implementation is described in detail below.
As shown in Figure 1, the power equipment infrared image multi-target orientation method based on deep learning, including following step
Suddenly：
Step 1：Standardized power equipment infrared image is obtained by substation equipment detection device；Wherein, pass through change
Power station technology personnel hand-held thermal imager collection in worksite picture or Intelligent Mobile Robot carry thermal infrared imager in inspection track
Photographs infrared image, specifically, the power equipment for including in the infrared image acquired has：Power transformer, is kept away breaker
Thunder device, voltage sensor, current sensor.
Step 2：Power equipment infrared image sample database is established, extraction training set is tested, demonstrate,proves collection and test set；Wherein, it will adopt
The power equipment infrared image collected is divided into training set, verification collection and test set according to a certain percentage, and to training set and verification
The infrared image of concentration adds label, is fabricated to the data set of PASCAL VOC formats, the sample image requirement of test set cannot
It is Chong Die with the image that verification is concentrated with training set, and label need not be added.
Step 3：Establish FASTER-RCNN depth targets detection neural network；Using the training set of sample database to being established
FASTER-RCNN depth targets detection neural network be trained, and pass through verify the set pair analysis model over-fitting degree carry out
Verification constantly obtains the connection weight and offset parameter of network model after debugging.As shown in Fig. 2, being FASTER-RCNN depth mesh
The basic block diagram of mark detection neural network, establishment step are as follows：
Step 3.1：Characteristics of image is extracted by depth convolutional neural networks (CNN) and generates characteristic spectrum.Referring to Fig. 3 institutes
Show, is the basic structure schematic diagram of a two-dimensional convolution neural network：All include a certain number of features in each layer of convolutional layer
Detector, after input is passed to convolutional layer, each characteristic detector can translate along image slide and carry out convolution algorithm, production
Raw corresponding characteristic spectrum.And with the stacking of convolutional layer, the feature of the deeper implied in input will be extracted simultaneously
Study.By taking the Feature capturing device of 3x3 as an example, input matrix obtains characteristic spectrum after the translation convolution of Feature capturing device.
I.e. for input matrix element aij, characteristic spectrum M is：
In formula, σ is expressed as the activation primitive of convolutional layer, ωklIt is the coefficient matrix of a 3x3, aiJ is the input of convolutional layer
I indicates that row j indicates row, and b is bias term.Each coefficient matrix can only learn a kind of single feature, therefore be needed in one layer more
A Feature capturing device, and the profound feature in data then needs to stack more convolutional layers to capture.
Step 3.2：Candidate region is generated to each pixel in characteristic spectrum using region candidate network (RPN)
(target detection frame).Specifically, RPN carries out the characteristic spectrum generated in step 3.1 by using a small Window-Network
Sliding calculates, each pixel in characteristic spectrum generates k optional target detection frames, and to each optional target
The offset for judging and being returned as Bbox that detection block carries out foreground or background by SoftMax graders is given a mark；
Step 3.3：Characteristics of image is combined with selected candidate region (target detection frame) by the ponds ROI layer, is led to
It crosses the Bbox Returns Law candidate window is zoomed in and out and finely tuned, rejects too small and target detection frame beyond boundary, final choice
Go out optimal region, and then completes target recognition and classification.
Step 4：The network model established using training carries out multi-targets recognition, to red to the infrared image in test set
The region and position of electrical power mains equipment extract in outer collection of illustrative plates, and generate the recognition result of power equipment infrared image；Tool
Body is that the FASTER-RCNN depth targets for being built the infrared image input step 3 in test set detect neural network, right
The region and position of electrical power mains equipment extract and obtain the recognition result of power equipment infrared image in infrared image.
Power equipment infrared image multi-target orientation method of the present invention based on deep learning forms FASTER-
The convolutional neural networks of RCNN carry out pre-training using Imagenet training sets, and corresponding convolutional layer activation primitive is ReLU,
Its formula is as follows：
F (x)=max (0, x)
In formula：When input signal is less than 0, it is 0 to export；When input signal is more than 0, output is equal to input.
Power equipment infrared image multi-target orientation method of the present invention based on deep learning, the ladder that model uses
Degree optimization algorithm is adaptability moments estimation algorithm (Adam), and in an iterative process, the variation of observation loss function value judges convergence
Situation, regularized learning algorithm rate.
The over-fitting in training process is reduced using dropout methods in model.Specifically, in training process
In, dropout technologies by according to certain probability by the hidden neuron random drop in network, i.e., by the input of the neuron
With output zero setting, the quantity of inner parameter in model on the one hand can be efficiently reduced, on the other hand also corresponds in a disguised form increase
The diversity of mode input data, to alleviate over-fitting to a certain extent.
Compared with prior art, the power equipment infrared image multi-target orientation method of the present invention based on deep learning,
Depth characteristic excavation is carried out to input infrared image by using convolutional neural networks, do not depend on manual feature extraction parameter,
And the over-fitting in training process is reduced by being introduced into dropout technologies, it is each so as to efficiently identify and position
The region and position of class electrical power mains equipment, improve the accuracy of positioning.And reduce to a certain extent to manual identified
Dependence, reduce manual labor amount, be worthy to be popularized.
Embodiment described above is only the preferred embodiments of the invention, and but not intended to limit the scope of the present invention, therefore
Change made by all shapes according to the present invention, principle, should all cover within the scope of the present invention.
Claims (7)
1. a kind of power equipment infrared image multi-target orientation method based on deep learning, which is characterized in that including following step
Suddenly：
Step 1：Standardized power equipment infrared image is obtained by substation equipment detection device；
Step 2：Establish power equipment infrared image sample database, extraction training set, verification collection and test set；
Step 3：FASTER-RCNN depth targets detection neural network is established, the training set using sample database is to being established
FASTER-RCNN depth targets detection neural network is trained, and the over-fitting degree by verifying the set pair analysis model is tested
Card constantly obtains the connection weight and offset parameter of network model after debugging；
Step 4：The network model established using training carries out multi-targets recognition, to infrared figure to the infrared image in test set
The region and position of trunk power equipment extract in spectrum, and generate the recognition result of power equipment infrared image.
2. a kind of power equipment infrared image multi-target orientation method based on deep learning according to claim 1,
It is characterized in that：In step 1, pass through substation technical staff hand-held thermal imager collection in worksite picture or Intelligent Mobile Robot
Thermal infrared imager is carried in inspection track photographs infrared image, wherein the power equipment for including in the infrared image acquired
Have：Power transformer, breaker, arrester, voltage sensor, current sensor.
3. a kind of power equipment infrared image multi-target orientation method based on deep learning according to claim 1,
It is characterized in that：In step 2, by collected power equipment infrared image according to preset ratio be divided into training set, verification collection and
Test set, and the infrared image concentrated to training set and verification adds label, is fabricated to the data set of PASCAL VOC formats,
In, the sample image of test set requires cannot be Chong Die with the image that verification is concentrated with training set, and need not add label.
4. a kind of power equipment infrared image multi-target orientation method based on deep learning according to claim 1,
It is characterized in that：In step 3, the FASTER-RCNN depth targets established detect neural network：For extracting image spy
The convolutional neural networks of characteristic spectrum are levied and generate, the area for generating candidate region to each pixel in characteristic spectrum
Domain candidate network, and interest pool area layer that characteristics of image is combined with selected candidate region；Finally by frame
The Return Law selects optimal region, and then completes target recognition and classification.
5. a kind of power equipment infrared image multi-target orientation method based on deep learning according to claim 1,
It is characterized in that：In step 4, the FASTER-RCNN depth targets infrared image input step 3 in test set built are examined
Neural network is surveyed, the region and position of electrical power mains equipment in infrared image are extracted and obtain power equipment infrared image
Recognition result.
6. a kind of power equipment infrared image multi-target orientation method based on deep learning according to claim 4,
It is characterized in that：The convolutional neural networks structure for forming FASTER-RCNN carries out pre-training by Imagenet training sets, corresponding
Convolutional layer activation primitive is linear amending unit function.
7. a kind of power equipment infrared image multi-target orientation method based on deep learning according to claim 4,
It is characterized in that：The gradient optimal method that the FASTER-RCNN depth targets detection neural network uses is adaptability moments estimation
Algorithm, in an iterative process, the variation of observation loss function value judge convergent, regularized learning algorithm rate, and in the network
Over-fitting in training process is reduced using dropout methods.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810199434.7A CN108564565A (en) | 2018-03-12 | 2018-03-12 | A kind of power equipment infrared image multi-target orientation method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810199434.7A CN108564565A (en) | 2018-03-12 | 2018-03-12 | A kind of power equipment infrared image multi-target orientation method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108564565A true CN108564565A (en) | 2018-09-21 |
Family
ID=63532700
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810199434.7A Pending CN108564565A (en) | 2018-03-12 | 2018-03-12 | A kind of power equipment infrared image multi-target orientation method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108564565A (en) |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109166336A (en) * | 2018-10-19 | 2019-01-08 | 福建工程学院 | A kind of real-time road condition information acquisition method for pushing based on block chain technology |
| CN109448061A (en) * | 2018-10-09 | 2019-03-08 | 西北工业大学 | A kind of underwater binocular visual positioning method without camera calibration |
| CN109522946A (en) * | 2018-10-31 | 2019-03-26 | 咪咕文化科技有限公司 | A kind of image classification model treatment method, apparatus and storage medium |
| CN109583344A (en) * | 2018-11-21 | 2019-04-05 | 红相股份有限公司 | A kind of power equipment Infrared Image Processing Method based on deep learning |
| CN109948613A (en) * | 2019-03-22 | 2019-06-28 | 国网重庆市电力公司电力科学研究院 | A kind of Infrared image recognition and device of arrester |
| CN110261703A (en) * | 2019-07-08 | 2019-09-20 | 厦门理工学院 | A kind of transformer fault method for early warning, terminal device and storage medium |
| CN110276394A (en) * | 2019-06-21 | 2019-09-24 | 扬州大学 | Power equipment classification method based on deep learning under a kind of small sample |
| CN110411580A (en) * | 2019-08-05 | 2019-11-05 | 国网湖南省电力有限公司 | A kind of diagnostic method and diagnostic system of equipment heating defect |
| CN110850723A (en) * | 2019-12-02 | 2020-02-28 | 西安科技大学 | Fault diagnosis and positioning method based on transformer substation inspection robot system |
| CN111209958A (en) * | 2020-01-03 | 2020-05-29 | 南方电网科学研究院有限责任公司 | Transformer substation equipment detection method and device based on deep learning |
| CN111260876A (en) * | 2018-11-30 | 2020-06-09 | 北京欣奕华科技有限公司 | Image processing method and device |
| CN111353487A (en) * | 2019-08-20 | 2020-06-30 | 国网浙江省电力有限公司台州供电公司 | Equipment information extraction method for transformer substation |
| CN111488918A (en) * | 2020-03-20 | 2020-08-04 | 天津大学 | Transformer substation infrared image equipment detection method based on convolutional neural network |
| CN111639579A (en) * | 2020-05-25 | 2020-09-08 | 南京航空航天大学 | Intelligent oil level identification method for power transformer bushing based on infrared image |
| CN111751762A (en) * | 2020-05-29 | 2020-10-09 | 江苏方天电力技术有限公司 | Power transformer state identification and safety prevention and control system |
| CN111764963A (en) * | 2020-07-06 | 2020-10-13 | 中国矿业大学(北京) | Rock burst prediction method based on fast-RCNN |
| CN112036410A (en) * | 2020-08-13 | 2020-12-04 | 华乘电气科技股份有限公司 | Electric power equipment infrared image segmentation method based on deep learning and image gradient |
| CN112200214A (en) * | 2020-08-24 | 2021-01-08 | 贵州大学 | PMSM multi-fault diagnosis method based on image recognition and convolutional neural network |
| CN112380952A (en) * | 2020-11-10 | 2021-02-19 | 广西大学 | Power equipment infrared image real-time detection and identification method based on artificial intelligence |
| CN112508073A (en) * | 2020-12-01 | 2021-03-16 | 杭州天铂云科光电科技有限公司 | Electric equipment infrared chart identification method combining deep learning with traditional algorithm |
| CN112598049A (en) * | 2020-12-18 | 2021-04-02 | 上海大学 | Target detection method for infrared image of buried object based on deep learning |
| CN112697798A (en) * | 2020-12-07 | 2021-04-23 | 国网信息通信产业集团有限公司 | Infrared image-oriented diagnosis method and device for current-induced thermal defects of power transformation equipment |
| CN114120160A (en) * | 2022-01-25 | 2022-03-01 | 成都合能创越软件有限公司 | Object space distinguishing method and device based on fast-RCNN, computer equipment and storage medium |
| CN114199381A (en) * | 2021-12-13 | 2022-03-18 | 国网陕西省电力公司延安供电公司 | Electrical equipment fault detection method for improving infrared detection model |
| EP3706268B1 (en) * | 2019-03-07 | 2022-06-29 | ABB Schweiz AG | Artificial intelligence monitoring system using infrared images to identify hotspots in a switchgear |
| CN115063418A (en) * | 2022-08-10 | 2022-09-16 | 北京航空航天大学 | Power battery temperature detection method based on image recognition |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103247039A (en) * | 2013-05-09 | 2013-08-14 | 河海大学常州校区 | Charged detection method of high-voltage cable based on composite vision |
| CN107271925A (en) * | 2017-06-26 | 2017-10-20 | 湘潭大学 | The level converter Fault Locating Method of modularization five based on depth convolutional network |
| CN107480730A (en) * | 2017-09-05 | 2017-12-15 | 广州供电局有限公司 | Power equipment identification model construction method and system, the recognition methods of power equipment |
-
2018
- 2018-03-12 CN CN201810199434.7A patent/CN108564565A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103247039A (en) * | 2013-05-09 | 2013-08-14 | 河海大学常州校区 | Charged detection method of high-voltage cable based on composite vision |
| CN107271925A (en) * | 2017-06-26 | 2017-10-20 | 湘潭大学 | The level converter Fault Locating Method of modularization five based on depth convolutional network |
| CN107480730A (en) * | 2017-09-05 | 2017-12-15 | 广州供电局有限公司 | Power equipment identification model construction method and system, the recognition methods of power equipment |
Cited By (36)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109448061A (en) * | 2018-10-09 | 2019-03-08 | 西北工业大学 | A kind of underwater binocular visual positioning method without camera calibration |
| CN109166336A (en) * | 2018-10-19 | 2019-01-08 | 福建工程学院 | A kind of real-time road condition information acquisition method for pushing based on block chain technology |
| CN109166336B (en) * | 2018-10-19 | 2020-08-07 | 福建工程学院 | Real-time road condition information acquisition and pushing method based on block chain technology |
| CN109522946A (en) * | 2018-10-31 | 2019-03-26 | 咪咕文化科技有限公司 | A kind of image classification model treatment method, apparatus and storage medium |
| CN109583344A (en) * | 2018-11-21 | 2019-04-05 | 红相股份有限公司 | A kind of power equipment Infrared Image Processing Method based on deep learning |
| CN109583344B (en) * | 2018-11-21 | 2019-11-19 | 红相股份有限公司 | A kind of power equipment Infrared Image Processing Method based on deep learning |
| CN111260876A (en) * | 2018-11-30 | 2020-06-09 | 北京欣奕华科技有限公司 | Image processing method and device |
| EP3706268B1 (en) * | 2019-03-07 | 2022-06-29 | ABB Schweiz AG | Artificial intelligence monitoring system using infrared images to identify hotspots in a switchgear |
| CN109948613A (en) * | 2019-03-22 | 2019-06-28 | 国网重庆市电力公司电力科学研究院 | A kind of Infrared image recognition and device of arrester |
| CN110276394A (en) * | 2019-06-21 | 2019-09-24 | 扬州大学 | Power equipment classification method based on deep learning under a kind of small sample |
| CN110261703A (en) * | 2019-07-08 | 2019-09-20 | 厦门理工学院 | A kind of transformer fault method for early warning, terminal device and storage medium |
| CN110411580A (en) * | 2019-08-05 | 2019-11-05 | 国网湖南省电力有限公司 | A kind of diagnostic method and diagnostic system of equipment heating defect |
| CN111353487A (en) * | 2019-08-20 | 2020-06-30 | 国网浙江省电力有限公司台州供电公司 | Equipment information extraction method for transformer substation |
| CN110850723A (en) * | 2019-12-02 | 2020-02-28 | 西安科技大学 | Fault diagnosis and positioning method based on transformer substation inspection robot system |
| CN111209958A (en) * | 2020-01-03 | 2020-05-29 | 南方电网科学研究院有限责任公司 | Transformer substation equipment detection method and device based on deep learning |
| CN111209958B (en) * | 2020-01-03 | 2023-11-14 | 南方电网科学研究院有限责任公司 | Substation equipment detection method and device based on deep learning |
| CN111488918A (en) * | 2020-03-20 | 2020-08-04 | 天津大学 | Transformer substation infrared image equipment detection method based on convolutional neural network |
| CN111639579A (en) * | 2020-05-25 | 2020-09-08 | 南京航空航天大学 | Intelligent oil level identification method for power transformer bushing based on infrared image |
| CN111639579B (en) * | 2020-05-25 | 2023-08-15 | 南京航空航天大学 | Intelligent oil level identification method for power transformer bushing based on infrared image |
| CN111751762A (en) * | 2020-05-29 | 2020-10-09 | 江苏方天电力技术有限公司 | Power transformer state identification and safety prevention and control system |
| CN111764963A (en) * | 2020-07-06 | 2020-10-13 | 中国矿业大学(北京) | Rock burst prediction method based on fast-RCNN |
| CN112036410A (en) * | 2020-08-13 | 2020-12-04 | 华乘电气科技股份有限公司 | Electric power equipment infrared image segmentation method based on deep learning and image gradient |
| CN112036410B (en) * | 2020-08-13 | 2022-06-07 | 华乘电气科技股份有限公司 | Electric power equipment infrared image segmentation method based on deep learning and image gradient |
| CN112200214A (en) * | 2020-08-24 | 2021-01-08 | 贵州大学 | PMSM multi-fault diagnosis method based on image recognition and convolutional neural network |
| CN112200214B (en) * | 2020-08-24 | 2022-09-13 | 贵州大学 | PMSM multi-fault diagnosis method based on image recognition and convolutional neural network |
| CN112380952A (en) * | 2020-11-10 | 2021-02-19 | 广西大学 | Power equipment infrared image real-time detection and identification method based on artificial intelligence |
| CN112380952B (en) * | 2020-11-10 | 2022-10-11 | 广西大学 | Power equipment infrared image real-time detection and identification method based on artificial intelligence |
| CN112508073A (en) * | 2020-12-01 | 2021-03-16 | 杭州天铂云科光电科技有限公司 | Electric equipment infrared chart identification method combining deep learning with traditional algorithm |
| CN112697798B (en) * | 2020-12-07 | 2023-03-31 | 国网信息通信产业集团有限公司 | Infrared image-oriented diagnosis method and device for current-induced thermal defects of power transformation equipment |
| CN112697798A (en) * | 2020-12-07 | 2021-04-23 | 国网信息通信产业集团有限公司 | Infrared image-oriented diagnosis method and device for current-induced thermal defects of power transformation equipment |
| CN112598049A (en) * | 2020-12-18 | 2021-04-02 | 上海大学 | Target detection method for infrared image of buried object based on deep learning |
| CN114199381A (en) * | 2021-12-13 | 2022-03-18 | 国网陕西省电力公司延安供电公司 | Electrical equipment fault detection method for improving infrared detection model |
| CN114120160B (en) * | 2022-01-25 | 2022-04-29 | 成都合能创越软件有限公司 | Object space distinguishing method and device based on fast-RCNN, computer equipment and storage medium |
| CN114120160A (en) * | 2022-01-25 | 2022-03-01 | 成都合能创越软件有限公司 | Object space distinguishing method and device based on fast-RCNN, computer equipment and storage medium |
| CN115063418A (en) * | 2022-08-10 | 2022-09-16 | 北京航空航天大学 | Power battery temperature detection method based on image recognition |
| CN115063418B (en) * | 2022-08-10 | 2022-11-01 | 北京航空航天大学 | Power battery temperature detection method based on image recognition |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108564565A (en) | A kind of power equipment infrared image multi-target orientation method based on deep learning | |
| CN110598736B (en) | Power equipment infrared image fault positioning, identifying and predicting method | |
| CN107944396B (en) | Knife switch state identification method based on improved deep learning | |
| CN103927534B (en) | A kind of online visible detection method of coding character based on convolutional neural networks | |
| CN106778604B (en) | Pedestrian re-identification method based on matching convolutional neural network | |
| CN108520226B (en) | Pedestrian re-identification method based on body decomposition and significance detection | |
| CN111784633B (en) | Insulator defect automatic detection algorithm for electric power inspection video | |
| CN105069779B (en) | A kind of architectural pottery surface detail pattern quality detection method | |
| CN107491781A (en) | A kind of crusing robot visible ray and infrared sensor data fusion method | |
| CN109118479A (en) | Defects of insulator identification positioning device and method based on capsule network | |
| CN109544501A (en) | A kind of transmission facility defect inspection method based on unmanned plane multi-source image characteristic matching | |
| CN112733950A (en) | Power equipment fault diagnosis method based on combination of image fusion and target detection | |
| CN111353487A (en) | Equipment information extraction method for transformer substation | |
| CN111814661A (en) | Human behavior identification method based on residual error-recurrent neural network | |
| CN108875819B (en) | Object and component joint detection method based on long-term and short-term memory network | |
| CN113378792B (en) | Weak supervision cervical cell image analysis method fusing global and local information | |
| CN107067410A (en) | A kind of manifold regularization correlation filtering method for tracking target based on augmented sample | |
| CN106326932A (en) | Power line inspection image automatic identification method based on neural network and power line inspection image automatic identification device thereof | |
| CN109544546A (en) | A kind of thermal-induced imagery defect characteristic extracting method based on multiple-objection optimization | |
| Xu et al. | D4Net: De-deformation defect detection network for non-rigid products with large patterns | |
| CN114494185B (en) | Electrical equipment fault detection method based on RGB-T multi-scale feature fusion | |
| CN109598711A (en) | A kind of thermal image defect extracting method based on feature mining and neural network | |
| CN115908354A (en) | Photovoltaic panel defect detection method based on double-scale strategy and improved YOLOV5 network | |
| CN113076816A (en) | Solar photovoltaic module hot spot identification method based on infrared and visible light images | |
| CN111402249A (en) | Image evolution analysis method based on deep learning |
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: 20180921 |
|
| RJ01 | Rejection of invention patent application after publication |