CN112307851A - Method and system for identifying bird nest on electric power iron tower - Google Patents
Method and system for identifying bird nest on electric power iron tower Download PDFInfo
- Publication number
- CN112307851A CN112307851A CN201910710454.0A CN201910710454A CN112307851A CN 112307851 A CN112307851 A CN 112307851A CN 201910710454 A CN201910710454 A CN 201910710454A CN 112307851 A CN112307851 A CN 112307851A
- Authority
- CN
- China
- Prior art keywords
- iron tower
- electric power
- power iron
- dimensional
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace 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
- G06N3/045—Combinations of networks
Abstract
The invention discloses a method for identifying a bird nest on an electric iron tower, which comprises a training step and an identification step, wherein the training step comprises the following steps: s100: collecting two-dimensional case images of the electric power iron tower; s200: constructing a convolutional neural network and training the convolutional neural network so as to simplify the data of the convolutional neural network; s300: constructing a depth belief network formed by stacking a plurality of limiting Boltzmann machines, reducing the dimension of two-dimensional data to one-dimensional data containing the image characteristics of the electric power iron tower, inputting the one-dimensional data into the depth belief network, and training the depth belief network by adopting the one-dimensional data so that the depth belief network outputs a recognition result; the identification step comprises: d100: inputting a two-dimensional image of the power iron tower to be identified into a trained convolutional neural network, and outputting two-dimensional data simplified by data by the convolutional neural network; d200: reducing the dimension of the two-dimensional data to one-dimensional data and inputting the one-dimensional data into a trained deep belief network; d300: and outputting the recognition result by the deep belief network.
Description
Technical Field
The present invention relates to a method and a system for recognizing in an electric power system, and more particularly, to a method and a system for recognizing an image in an electric power system device.
Background
With the continuous increase of high-voltage transmission lines and the gradual improvement of ecological environment in China, bird accidents obviously rise, the loss caused by the bird accidents also obviously increases, and the serious threat is brought to the safe and stable operation of a power grid.
According to statistics, the line fault caused by bird activity is only second to lightning damage and external force damage, and occupies the 3 rd position of the total number of line faults. The iron tower has high bird damage failure rate, and the main reasons are that the iron tower is large and stable, and the visual field of birds is wide, so that the iron tower is suitable for birds to stay and nest.
Line trip accidents caused by bird damage on high voltage towers account for a considerable proportion. The main causes of line tripping are bird droppings flashover, bird nest material short circuits, and bird body short circuits. Wherein short circuits of bird nest material and thus of other types of bird damage account for a large proportion. If a corresponding method can be found, and workers are sent to remove the bird nest after the bird nest is effectively detected, the proportion of bird damage can be greatly reduced.
The bird nest is generally built by branches and hay, the nondirectional property and the irregularity of the bird nest and the shielding of steel destroy the local texture of the bird nest, and the situation that the bird nest characteristic cannot be well represented if only a single texture characteristic is considered is determined. Therefore, if a single color feature or a single texture feature is adopted to detect the bird nest on the iron tower, missing detection and false detection can occur.
Based on this, it is desirable to obtain a method for identifying a bird nest on an electric power iron tower, which can identify whether the electric power iron tower has the bird nest through the acquired image, maintain the running stability of electric power equipment in a more convenient and efficient manner, and reduce the bird damage failure rate of the iron tower.
Disclosure of Invention
One of the objectives of the present invention is to provide a method for identifying a bird nest on an electric power iron tower, which can identify whether the electric power iron tower has the bird nest through an acquired image, so as to maintain the operation stability of electric power equipment in a more convenient and efficient manner, and reduce the bird damage failure rate of the iron tower.
According to the above object, the present invention provides a method for identifying a bird nest on an electric power iron tower, which includes a training step and an identification step, wherein the training step includes:
s100: collecting two-dimensional case images of the electric power iron tower;
s200: constructing a convolutional neural network and training the convolutional neural network so that the convolutional neural network carries out data simplification processing on the two-dimensional case image of the power iron tower and outputs two-dimensional data which are subjected to data simplification and contain the image characteristics of the power iron tower;
s300: constructing a depth belief network formed by stacking a plurality of limiting Boltzmann machines, reducing the dimension of the two-dimensional data to one-dimensional data containing the image characteristics of the electric power iron tower, inputting the one-dimensional data into the depth belief network, and training the depth belief network by adopting the one-dimensional data so as to enable the depth belief network to output the recognition result whether the electric power iron tower has the bird nest or not;
the identification step comprises:
d100: inputting a two-dimensional image of the power iron tower to be identified into a trained convolutional neural network, and outputting two-dimensional data simplified by data by the convolutional neural network;
d200: reducing the dimension of the two-dimensional data to one-dimensional data and inputting the one-dimensional data into a trained deep belief network;
d300: and the deep belief network outputs the identification result of whether the electric iron tower has the bird nest or not.
It should be noted that, in the technical solution of the present invention, the two-dimensional case image of the power tower can be obtained by shooting through an image acquisition device, such as an unmanned aerial vehicle with a camera device. In addition, considering that the ground conditions are complex, especially when various branches, bushes, thatch piles and the like exist in the field, if the shooting is performed from bottom to top (or from bottom to oblique top direction), most of the background patterns are sky, and the former background patterns are much simpler than those from top to bottom (including oblique bottom direction); in addition, considering that the electric power line can be subjected to green cutting before construction, namely trees in a certain range along the electric power line can be cut off, the unmanned aerial vehicle can preferably fly at a certain distance below the electric power line, especially below the cross arm area of the electric power iron tower, because the cross arm area of the electric power iron tower is high in ground clearance, good in visual field and large in cross arm area, and is suitable for building a bird nest, and the method of mounting a wide-angle camera at the top of the unmanned aerial vehicle is used for shooting images of the electric power iron tower.
Further, in the identification method of bird's nests on the power iron tower, the convolutional neural network at least comprises five convolutional layers and five pooling layers.
Further, in the identification method of bird nests on the electric power iron tower, the convolutional neural network further comprises two full connection layers.
Further, in the identification method of bird nests on the electric power iron tower, each convolution layer comprises a plurality of convolution units, and the parameters of each convolution unit are obtained through optimization of a back propagation algorithm.
Further, in the identification method of bird's nests on an electric power tower according to the present invention, the pooling layer is configured to reduce the size of the input electric power tower two-dimensional case image and the electric power tower two-dimensional image to be identified without changing the number thereof.
Further, in the identification method of bird nests on the electric power iron tower, the pooling layer adopts an average pooling method.
Further, in the method for identifying the bird nest on the electric power iron tower, the output of the last pooling layer of the convolutional neural network is connected with the deep belief network.
Further, in the identification method of the bird nest on the power iron tower, the identification result includes "bird nest present", "bird nest absent", and "indeterminate".
Further, in the identification method of bird's nests on the electric power iron tower according to the present invention, the image features of the electric power iron tower include color features, texture features, and shape features.
Further, in the identification method of the bird nest on the electric power iron tower, the camera device arranged on the unmanned aerial vehicle is adopted to collect the two-dimensional case image of the electric power iron tower and the two-dimensional image of the electric power iron tower to be identified.
Accordingly, another object of the present invention is to provide a system for identifying a bird nest on an electric power iron tower, which can identify whether the electric power iron tower has the bird nest through collected images, so as to maintain the stability of the operation of the electric power equipment in a more convenient and efficient manner and reduce the bird damage failure rate of the iron tower.
According to the above purpose, the present invention provides a system for identifying a bird nest on an electric power tower, which includes a data acquisition module and a data processing module connected by data, and the identification system implements the above identification method for a bird nest on an electric power tower.
Compared with the prior art, the method and the system for identifying the bird nest on the electric iron tower have the following advantages and beneficial effects:
the method for identifying the bird nest on the electric power iron tower can identify whether the electric power iron tower has the bird nest or not through the acquired image, maintain the running stability of the electric power equipment in a more convenient and efficient mode and reduce the bird damage fault rate of the iron tower.
In addition, the identification system of the bird nest on the electric power iron tower also has the advantages and beneficial effects as described above.
Drawings
Fig. 1 is a schematic structural diagram of a convolutional neural network used in an embodiment of the method for identifying a bird nest on an electric iron tower according to the present invention.
Fig. 2 is a schematic structural diagram of a limiting boltzmann machine used in an embodiment of the method for identifying a bird nest on an electric iron tower according to the present invention.
Fig. 3 is a schematic structural diagram of a deep belief network used in an embodiment of the method for identifying a bird nest on an electric power iron tower according to the present invention.
Fig. 4 illustrates a training process of a deep belief network used in an embodiment of the identification method of bird nests on an electric power iron tower according to the present invention.
Fig. 5 illustrates a connection structure of a convolutional neural network and a deep belief network used in an embodiment of the identification method of bird nests on an electric power iron tower according to the present invention.
Detailed Description
The method and system for identifying bird's nests on an electric power iron tower according to the present invention will be further explained and illustrated with reference to the drawings and the specific embodiments of the present invention, however, the explanation and the illustration do not unduly limit the technical solution of the present invention.
In this embodiment, the identification system of bird's nest on electric power iron tower includes data connection's data acquisition module and data processing module, and wherein, data acquisition module can adopt the method of installing wide-angle camera at the unmanned aerial vehicle top, shoots electric power iron tower image with from the mode up down in the oblique below of power line.
And the data processing module performs training and recognition processing after acquiring the shot images through data transmission.
The identification system executes the following training steps and identification steps, wherein the training steps comprise:
s100: collecting two-dimensional case images of the electric power iron tower;
s200: constructing a convolutional neural network and training the convolutional neural network so that the convolutional neural network carries out data simplification processing on the two-dimensional case image of the power iron tower and outputs two-dimensional data which are subjected to data simplification and contain the image characteristics of the power iron tower;
s300: constructing a depth belief network formed by stacking a plurality of limiting Boltzmann machines, reducing the dimension of the two-dimensional data to one-dimensional data containing the image characteristics of the electric power iron tower, inputting the one-dimensional data into the depth belief network, and training the depth belief network by adopting the one-dimensional data so as to enable the depth belief network to output the recognition result whether the electric power iron tower has the bird nest or not;
and the step of identifying comprises:
d100: inputting a two-dimensional image of the power iron tower to be identified into a trained convolutional neural network, and outputting two-dimensional data simplified by data by the convolutional neural network;
d200: reducing the dimension of the two-dimensional data to one-dimensional data and inputting the one-dimensional data into a trained deep belief network;
d300: and the deep belief network outputs the identification result of whether the electric iron tower has the bird nest or not.
In this embodiment, the image features of the power tower include color features, texture features, and shape features.
The structure of the convolutional neural network can be seen in fig. 1. Fig. 1 is a schematic structural diagram of a convolutional neural network used in an embodiment of the method for identifying a bird nest on an electric iron tower according to the present invention.
As shown in fig. 1, in the present embodiment, the convolutional neural network (CNN for short) includes at least five convolutional layers and five pooling layers. In addition, in this embodiment, the convolutional neural network further includes two fully-connected layers. The parameters used are as follows:
the relevant parameters are used as follows:
layer 1:
inputting: length x width x channel (224 x 3)
And (3) rolling layers: length X width X channel X step (11X 96X 4)
A pooling layer: maximum pooling length × width × step size (3 × 3 × 2)
And (3) outputting: length x width x channel (27 x 96)
Layer 2:
inputting: length x width x channel (27 x 96)
And (3) rolling layers: length × width × channel × step (5 × 5 × 128 × 1), and all zero padding is employed
A pooling layer: maximum pooling length × width × step size (3 × 3 × 2)
And (3) outputting: length x width x channel (13 x 256)
Layer 3:
inputting: length x width x channel (13 x 256)
And (3) rolling layers: length × width × channel × step (3 × 3 × 192 × 1), and all zero padding is employed
A pooling layer: maximum pooling length × width × step size (3 × 3 × 2)
And (3) outputting: length x width x channel (13 x 384)
Layer 4:
inputting: length x width x channel (13 x 384)
And (3) rolling layers: length × width × channel × step (3 × 3 × 192 × 1), and all zero padding is employed
A pooling layer: maximum pooling length × width × step size (3 × 3 × 2)
And (3) outputting: length x width x channel (13 x 384)
Layer 5:
an input unit: length x width x channel (13 x 384)
And (3) rolling layers: length × width × channel × step (3 × 3 × 128 × 1), and all zero padding is employed
A pooling layer: maximum pooling length × width × step size (3 × 3 × 2)
And (3) outputting: length x width x channel (6 x 256)
Number of nodes of the whole chain layer: [9216,27648,3]
It should be noted that in the present disclosure, the full-chain layer of the CNN only plays a role in the initial individual training of the CNN, and when interfacing with a deep belief network (DBN for short), the output of the pooling layer of the layer 5 is adopted to interface with the DBN. Therefore, the number of nodes where the CNN and the DBN are connected is 9216 nodes.
In addition, it should be noted that each convolution layer is composed of a plurality of convolution units, parameters of each convolution unit are obtained through optimization of a back propagation algorithm, and different input features can be extracted and obtained through convolution operation.
While the pooling layer may reduce the sensitivity of the feature map output by local averaging and downsampling. In the present embodiment, the pooling layer may preferably employ an average pooling method, which may reduce the size of the input electric power tower two-dimensional case images and the electric power tower two-dimensional images to be recognized without changing the number thereof.
Of course, in some other embodiments, the maximum pooling method may also be used to reduce the size of the input two-dimensional case images of the power tower and the two-dimensional images of the power tower to be identified without changing the number thereof.
In the scheme, a DBN formed by stacking a plurality of Restricted Boltzmann Machines (RBMs) is adopted for subsequent analysis of the iron tower image, because the RBMs have strong unsupervised learning capability and can learn complex rules in data.
Fig. 2 is a schematic structural diagram of a limiting boltzmann machine used in an embodiment of the method for identifying a bird nest on an electric iron tower according to the present invention. As shown in fig. 2, the connection between the visible layer h and the hidden layer v of the RBM is undirected, and there is no connection between any two visible units or two hidden units of the same layer.
Fig. 3 is a schematic structural diagram of a deep belief network used in an embodiment of the method for identifying a bird nest on an electric power iron tower according to the present invention.
As shown in FIG. 3, the DBN is a generative model formed by overlapping a plurality of RBMs, 3 RBMs (namely RBM1, RBM2 and RBM3) are schematically shown in FIG. 3, and when training, a data vector and a first layer hidden layer are taken as the RBM1, then parameters of the RBM1 are trained, and then h is taken as a parameter of the RBM11Viewed as a visual vector, take h2The RBM2 continues to be trained as an implicit vector, followed by the implicit layer of RBM2 as the visual layer of RBM3, and so on until training is complete. In addition, W in fig. 31、W2And W3Respectively represent RBM1,2. Weight between v-layer and h-layer in 3.
Because RBM same-layer units are not interconnected, the variables of the same-layer units are independent, and the distribution probability p (v | h) expression of each node is as follows:
wherein E represents an energy function, x is a random variable, Z is a normalization factor, and the Z expression is as follows:
let n and m be the number of nodes of input layer v and hidden layer h, where vi、hjRespectively representing the ith and j node states of the input layer v and the hidden layer h, the energy function E (v, h) of a given state (v, h) of the RBM is as follows:
Wijis a node vi、hjConnection weight of ai、bjDenotes the offset of the i-th and j-th nodes, respectively, where θ ═ Wij,ai,bjIt is the RBM model parameter.
And for RBM training, the method can be completed by alternately sampling visible units and hidden units through a learning algorithm of contrast divergence, as shown in the following formula, wherein n and m are the unit numbers of the hidden layer and the visible layer:
wherein P (h | v) and P (v | h) are eachRepresenting conditional probabilities of implicit and visible units; p (h)j| v) indicates that h appears at jth node of h layer in the state of vjThe conditional probability of (a); p (v)j| h) indicates that v appears at jth node of v layer in h statejThe conditional probability of (2).
The above calculation process is known to those skilled in the art, and therefore, will not be described herein.
The training process of the deep belief network is divided into an unsupervised learning stage and a supervised fine tuning stage, the deep belief network unsupervised trains each layer of RBM network, parameters obtained by training each single layer are used as initial parameters of neurons of each layer of the deep network, feature vectors are ensured to be mapped to different feature spaces, feature information is kept as much as possible, and the process is pre-training. After the deep belief network trains the network parameters layer by layer, the deep network is reversely trained by using a BP algorithm, so that the network parameters are converged at a better position finally.
And performing parameter training updating by using a contrast divergence algorithm in the training process, wherein the parameter updating process is as follows:
Δci＝p(hi＝1|v(0))-p(hi＝1|v(k))
wherein, Δ wijFor updating of the weights, Δ bj、ΔcjRespectively updating the deviation of the visible layer node and the hidden layer node; p is the conditional probability; v. of(0)、v(k)V values for 0 th and k th generations, respectively; v. ofj (0)、vj (k)V of jth node of v layer of 0 th generation and k th generation respectivelyjThe value is obtained.
Fig. 4 illustrates a training process of a deep belief network used in an embodiment of the identification method of bird nests on an electric power iron tower according to the present invention.
As shown in fig. 4, in the DBN network training process, the state parameters of the training samples are input into the multiple layers of RBMs, and are updated layer by layer, and finally, the parameter optimization of the model is completed, that is, the training results of the RBMs and the characteristic input of the samples are reversely adjusted by using the BP algorithm, and finally, the DBN network model is obtained. In the embodiment shown in fig. 4, μmay take a value of 3%.
Fig. 5 illustrates a connection structure of a convolutional neural network and a deep belief network used in an embodiment of the identification method of bird nests on an electric power iron tower according to the present invention.
As shown in fig. 5, in the present embodiment, the first 5 layers of weights of the CNN are preliminarily optimized, and the CNN has good adaptability to changes of rotation, displacement and scaling of an image, so as to solve problems of related changes of the image and changes of image features due to local input transformation, and perform data simplification processing on the two-dimensional case image of the power tower, so that the CNN outputs two-dimensional data containing image features of the power tower, which is subjected to data simplification;
and then, connecting the output of the 5 th pooling layer with the DBN aiming at the initially trained CNN so as to enable the output to be directly input into the DBN, performing initial training on the DBN, and finally combining the CNN with the DBN to perform overall training to obtain the CNN-DBN composite deep learning network for bird nest image recognition.
And finally, inputting the two-dimensional image of the power tower to be recognized into the trained CNN and the trained DBN, and outputting a recognition result of whether the power tower has a bird nest or not through the DBN, wherein the recognition result comprises 'bird nest presence', 'no bird nest' and 'uncertainty'.
In summary, the method for identifying the bird nest on the electric power iron tower can identify whether the electric power iron tower has the bird nest or not through the acquired image, maintain the running stability of the electric power equipment in a more convenient and efficient manner, and reduce the bird damage fault rate of the iron tower.
In addition, the identification system of the bird nest on the electric power iron tower also has the advantages and beneficial effects as described above.
It should be noted that the prior art in the protection scope of the present invention is not limited to the examples given in the present application, and all the prior art which is not inconsistent with the technical scheme of the present invention, including but not limited to the prior patent documents, the prior publications and the like, can be included in the protection scope of the present invention.
In addition, the combination of the features in the present application is not limited to the combination described in the claims of the present application or the combination described in the embodiments, and all the features described in the present application may be freely combined or combined in any manner unless contradictory to each other.
It should also be noted that the above-mentioned embodiments are only specific embodiments of the present invention. It is apparent that the present invention is not limited to the above embodiments and similar changes or modifications can be easily made by those skilled in the art from the disclosure of the present invention and shall fall within the scope of the present invention.
Claims (11)
1. A method for identifying bird nests on an electric power iron tower is characterized by comprising a training step and an identification step, wherein the training step comprises the following steps:
s100: collecting two-dimensional case images of the electric power iron tower;
s200: constructing a convolutional neural network and training the convolutional neural network so that the convolutional neural network carries out data simplification processing on the two-dimensional case image of the power iron tower and outputs two-dimensional data which are subjected to data simplification and contain the image characteristics of the power iron tower;
s300: constructing a depth belief network formed by stacking a plurality of limiting Boltzmann machines, reducing the dimension of the two-dimensional data to one-dimensional data containing the image characteristics of the electric power iron tower, inputting the one-dimensional data into the depth belief network, and training the depth belief network by adopting the one-dimensional data so as to enable the depth belief network to output the recognition result whether the electric power iron tower has the bird nest or not;
the identifying step includes:
d100: inputting a two-dimensional image of the power iron tower to be identified into a trained convolutional neural network, and outputting two-dimensional data simplified by data by the convolutional neural network;
d200: reducing the dimension of the two-dimensional data to one-dimensional data and inputting the one-dimensional data into a trained deep belief network;
d300: and the deep belief network outputs the identification result of whether the electric iron tower has the bird nest or not.
2. The method according to claim 1, wherein the convolutional neural network comprises at least five convolutional layers and five pooling layers.
3. The method for identifying bird nests on a power tower of claim 2, wherein the convolutional neural network further comprises two fully connected layers.
4. The method for identifying bird nests on an electric power iron tower of claim 2, wherein each convolution layer comprises a plurality of convolution units, and parameters of each convolution unit are optimized through a back propagation algorithm.
5. The method of claim 2, wherein the pooling layer is configured to reduce the size of the input two-dimensional case image of the power tower and the two-dimensional image of the power tower to be identified without changing the number thereof.
6. The method for identifying bird nests on an electric power iron tower of claim 2, wherein the pooling layer employs an average pooling method.
7. The method for identifying bird nests on a power tower of claim 1, wherein an output of a last pooling layer of the convolutional neural network is connected to the deep belief network.
8. The method for identifying the bird nest on the power tower according to claim 1, wherein the identification result includes "bird nest present", "bird nest absent", and "indeterminate".
9. The method for identifying the bird nest on the power tower according to claim 1, wherein the image features of the power tower comprise color features, texture features and shape features.
10. The method for identifying the bird nest on the electric power iron tower as claimed in claim 1, wherein the two-dimensional case image of the electric power iron tower and the two-dimensional image of the electric power iron tower to be identified are collected by a camera device installed on the unmanned aerial vehicle.
11. An identification system for bird nests on an electric power iron tower, which comprises a data acquisition module and a data processing module which are connected by data, and implements the identification method for bird nests on the electric power iron tower according to any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910710454.0A CN112307851A (en) | 2019-08-02 | 2019-08-02 | Method and system for identifying bird nest on electric power iron tower |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910710454.0A CN112307851A (en) | 2019-08-02 | 2019-08-02 | Method and system for identifying bird nest on electric power iron tower |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112307851A true CN112307851A (en) | 2021-02-02 |
Family
ID=74485993
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910710454.0A Pending CN112307851A (en) | 2019-08-02 | 2019-08-02 | Method and system for identifying bird nest on electric power iron tower |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112307851A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104483326A (en) * | 2014-12-19 | 2015-04-01 | 长春工程学院 | High-voltage wire insulator defect detection method and high-voltage wire insulator defect detection system based on deep belief network |
| CN106971152A (en) * | 2017-03-16 | 2017-07-21 | 天津大学 | A kind of method of Bird's Nest in detection transmission line of electricity based on Aerial Images |
| CN106980827A (en) * | 2017-03-16 | 2017-07-25 | 天津大学 | A kind of method of Bird's Nest in identification transmission line of electricity based on Aerial Images |
| CN107392901A (en) * | 2017-07-24 | 2017-11-24 | 国网山东省电力公司信息通信公司 | A kind of method for transmission line part intelligence automatic identification |
| CN107449994A (en) * | 2017-07-04 | 2017-12-08 | 国网江苏省电力公司电力科学研究院 | Partial discharge method for diagnosing faults based on CNN DBN networks |
| CN108597053A (en) * | 2018-04-25 | 2018-09-28 | 北京御航智能科技有限公司 | Shaft tower and channel targets identification based on image data and neural network and defect diagnostic method |
| CN109086817A (en) * | 2018-07-25 | 2018-12-25 | 西安工程大学 | A kind of Fault Diagnosis for HV Circuit Breakers method based on deepness belief network |
| CN109164362A (en) * | 2018-09-25 | 2019-01-08 | 上海交通大学 | A kind of recognition methods and system of direct current cables shelf depreciation defect failure |
| CN109389180A (en) * | 2018-10-30 | 2019-02-26 | 国网四川省电力公司广元供电公司 | A power equipment image-recognizing method and inspection robot based on deep learning |
| CN109472229A (en) * | 2018-10-30 | 2019-03-15 | 福州大学 | Shaft tower Bird's Nest detection method based on deep learning |
| CN109765462A (en) * | 2019-03-05 | 2019-05-17 | 国家电网有限公司 | Fault detection method, device and the terminal device of transmission line of electricity |
-
2019
- 2019-08-02 CN CN201910710454.0A patent/CN112307851A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104483326A (en) * | 2014-12-19 | 2015-04-01 | 长春工程学院 | High-voltage wire insulator defect detection method and high-voltage wire insulator defect detection system based on deep belief network |
| CN106971152A (en) * | 2017-03-16 | 2017-07-21 | 天津大学 | A kind of method of Bird's Nest in detection transmission line of electricity based on Aerial Images |
| CN106980827A (en) * | 2017-03-16 | 2017-07-25 | 天津大学 | A kind of method of Bird's Nest in identification transmission line of electricity based on Aerial Images |
| CN107449994A (en) * | 2017-07-04 | 2017-12-08 | 国网江苏省电力公司电力科学研究院 | Partial discharge method for diagnosing faults based on CNN DBN networks |
| CN107392901A (en) * | 2017-07-24 | 2017-11-24 | 国网山东省电力公司信息通信公司 | A kind of method for transmission line part intelligence automatic identification |
| CN108597053A (en) * | 2018-04-25 | 2018-09-28 | 北京御航智能科技有限公司 | Shaft tower and channel targets identification based on image data and neural network and defect diagnostic method |
| CN109086817A (en) * | 2018-07-25 | 2018-12-25 | 西安工程大学 | A kind of Fault Diagnosis for HV Circuit Breakers method based on deepness belief network |
| CN109164362A (en) * | 2018-09-25 | 2019-01-08 | 上海交通大学 | A kind of recognition methods and system of direct current cables shelf depreciation defect failure |
| CN109389180A (en) * | 2018-10-30 | 2019-02-26 | 国网四川省电力公司广元供电公司 | A power equipment image-recognizing method and inspection robot based on deep learning |
| CN109472229A (en) * | 2018-10-30 | 2019-03-15 | 福州大学 | Shaft tower Bird's Nest detection method based on deep learning |
| CN109765462A (en) * | 2019-03-05 | 2019-05-17 | 国家电网有限公司 | Fault detection method, device and the terminal device of transmission line of electricity |
Non-Patent Citations (3)
| Title |
|---|
| 师飘: "输电线路上鸟巢的检测算法研究", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 * |
| 杨沛: "双判别器生成对抗网络及其在接触网鸟巢检测的应用研究", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 * |
| 王海涛等: "两级上下文卷积网络宽视场图像小目标检测方法", 《计算机测量与控制》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106356757A (en) | Method for inspecting electric power lines by aid of unmanned aerial vehicle on basis of human vision characteristics | |
| CN109886155B (en) | Single-plant rice detection and positioning method, system, equipment and medium based on deep learning | |
| CN106406346A (en) | Plan method for rapid coverage track search coordinated by multiple UAVs (Unmanned Aerial Vehicles) | |
| CN113435644B (en) | Emergency prediction method based on deep bidirectional long-short term memory neural network | |
| CN111191559B (en) | Overhead line early warning system obstacle recognition method based on time convolution neural network | |
| CN106373397A (en) | Fuzzy neural network-based remote sensing image road traffic situation analysis method | |
| TC et al. | Enhanced deep convolutional neural network for fault signal recognition in the power distribution system | |
| CN111860514A (en) | Orchard scene multi-class real-time segmentation method based on improved deep Lab | |
| CN116343330A (en) | Abnormal behavior identification method for infrared-visible light image fusion | |
| CN115620107A (en) | Transformer substation bird-involved fault related bird species identification method based on deep learning | |
| CN106338674A (en) | DC cable connector insulation fault diagnosis method and system based on improved ECOC classifier | |
| CN114137358B (en) | Power transmission line fault diagnosis method based on graph convolution neural network | |
| CN115116054A (en) | Insect pest identification method based on multi-scale lightweight network | |
| CN116245871A (en) | Power transmission line abnormal target detection method based on improved YOLOX algorithm | |
| CN112307851A (en) | Method and system for identifying bird nest on electric power iron tower | |
| Gai et al. | Cherry detection algorithm based on improved YOLOv5s network | |
| CN112215197A (en) | Underground cable fault early warning method and system based on artificial intelligence | |
| CN113988189B (en) | Migration fault diagnosis method for cross-wind turbine generator | |
| CN112817318B (en) | Multi-unmanned-boat collaborative search control method and system | |
| Hao et al. | Based on Surf feature extraction and insulator damage identification for capsule networks | |
| Hu et al. | Research on pest and disease recognition algorithms based on convolutional neural network | |
| CN112199980A (en) | Overhead line robot obstacle identification method | |
| Wang et al. | A novel algorithm for edge detection of remote sensing image based on CNN and PSO | |
| CN108573199B (en) | Method and system for judging disaster level of Chinese pine | |
| CN117876741A (en) | Point cloud classification method and system based on multidimensional feature fusion attention pooling network |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210202 |
|
| WD01 | Invention patent application deemed withdrawn after publication |